Professional Documents
Culture Documents
and
ALGEBRA NOTES
Preface
2. These Notes are not intended to be a substitute for attending lectures or tutorials. The
lectures will expand on the material in the notes and help you to understand it.
3. These Notes may seem to contain a lot of material but not all of this material is equally
important. One aim of the lectures will be to give you a clearer idea of the relative importance
of the topics covered in the Notes.
4. Use the tutorials for the purpose for which they are intended, that is, to ask questions about
both the theory and the problems being covered in the current lectures.
5. Some of the material in these Notes is more difficult than the rest. This extra material
is marked with the symbol [H]. Material marked with an [X] is intended for students in
MATH1141.
7. It is essential for you to do problems which are given at the end of each chapter. If you
find that you do not have time to attempt all of the problems, you should at least attempt a
representative selection of them. You will find advice about this on Moodle. You should also
work through the Online Tutorals that you will find on Moodle.
8. You will be expected to use the computer algebra package Maple in tests and understand
Maple syntax and output for the end of term examination.
Note.
These notes have been prepared by many members of the School of Mathematics and Statistics.
The main contributors include Peter Blennerhassett, Peter Brown, Shaun Disney, Peter Donovan,
c
2019 School of Mathematics and Statistics, UNSW Sydney
iv CHAPTER 0. PREFACE
Ian Doust, David Hunt, Chi Mak, Elvin Moore and Colin Sutherland. Copyright is vested in The
c
University of New South Wales,
2019.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CONTENTS v
Contents
Preface iii
Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Syllabus for MATH1131 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Syllabus for MATH1141 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
1 INTRODUCTION TO VECTORS 1
1.1 Vector quantities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Geometric vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Two dimensional vector quantities . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Vector quantities and Rn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.1 Vectors in R2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.2 Vectors in Rn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Rn and analytic geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.1 Two dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.2 Three dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.3 n-dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4 Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.4.1 Lines in R2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.4.2 Lines in R3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.4.3 Lines through two given points (in Rn ) . . . . . . . . . . . . . . . . . . . . . . 27
1.5 Planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.5.1 Linear combination and span . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.5.2 Parametric vector form of a plane . . . . . . . . . . . . . . . . . . . . . . . . 31
1.5.3 Cartesian form of a plane in R3 . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.6 Vectors and Maple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2 VECTOR GEOMETRY 45
2.1 Lengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.2 The dot product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.2.1 Arithmetic properties of the dot product . . . . . . . . . . . . . . . . . . . . . 48
2.2.2 Geometric interpretation of the dot product in Rn . . . . . . . . . . . . . . . 48
2.3 Applications: orthogonality and projection . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.1 Orthogonality of vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.2 Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
c
2019 School of Mathematics and Statistics, UNSW Sydney
vi CONTENTS
3 COMPLEX NUMBERS 81
3.1 A review of number systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.2 Introduction to complex numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.3 The rules of arithmetic for complex numbers . . . . . . . . . . . . . . . . . . . . . . 84
3.4 Real parts, imaginary parts and complex conjugates . . . . . . . . . . . . . . . . . . 86
3.5 The Argand diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.6 Polar form, modulus and argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.7 Properties and applications of the polar form . . . . . . . . . . . . . . . . . . . . . . 94
3.7.1 The arithmetic of polar forms . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.7.2 Powers of complex numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.7.3 Roots of complex numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.8 Trigonometric applications of complex numbers . . . . . . . . . . . . . . . . . . . . . 102
3.9 Geometric applications of complex numbers . . . . . . . . . . . . . . . . . . . . . . . 105
3.10 Complex polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.10.1 Roots and factors of polynomials . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.10.2 Factorisation of polynomials with real coefficients . . . . . . . . . . . . . . . . 112
3.11 Appendix: A note on proof by induction . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.12 Appendix: The Binomial Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.13 Complex numbers and Maple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
c
2019 School of Mathematics and Statistics, UNSW Sydney
CONTENTS vii
5 MATRICES 173
5.1 Matrix arithmetic and algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
5.1.1 Equality, addition and multiplication by a scalar . . . . . . . . . . . . . . . . 174
5.1.2 Matrix multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
5.1.3 Matrix arithmetic and systems of linear equations . . . . . . . . . . . . . . . 182
5.2 The transpose of a matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
5.2.1 Some uses of transposes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
5.2.2 Some properties of transposes . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
5.3 The inverse of a matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
5.3.1 Some useful properties of inverses . . . . . . . . . . . . . . . . . . . . . . . . . 189
5.3.2 Calculating the inverse of a matrix . . . . . . . . . . . . . . . . . . . . . . . . 190
5.3.3 Inverse of a 2 × 2 matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
5.3.5 Inverses and solution of Ax = b . . . . . . . . . . . . . . . . . . . . . . . . . . 195
5.4 Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
5.4.1 The definition of a determinant . . . . . . . . . . . . . . . . . . . . . . . . . . 196
5.4.2 Properties of determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
5.4.3 The efficient numerical evaluation of determinants . . . . . . . . . . . . . . . 201
5.4.4 Determinants and solutions of Ax = b . . . . . . . . . . . . . . . . . . . . . . 204
5.5 Matrices and Maple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
INDEX 240
c
2019 School of Mathematics and Statistics, UNSW Sydney
viii CONTENTS
c
2019 School of Mathematics and Statistics, UNSW Sydney
CONTENTS ix
The algebra course for both MATH1131 and MATH1141 is based on the MATH1131/MATH1141
Algebra Notes that are included in the Course Pack.
Please note that the order of the syllabus changed in 2014, in accordance with requests from
the Engineering Faculty and the School of Physics. It is important to note this in regard to the
class tests from previous years.
The computer package Maple will be used in the algebra course. An introduction to Maple is
included in the booklet Computing Laboratories Information and First Year Maple Notes .
The lecture timetable is given below. Lecturers will try to follow this timetable, but some vari-
ations may be unavoidable, especially in MATH1141 classes and lecture groups affected by public
holidays.
c
2019 School of Mathematics and Statistics, UNSW Sydney
x CONTENTS
Chapter 5. Matrices
Lecture 21. Operations on matrices. Transposes. (Sections 5.1, 5.2).
Lecture 22. Inverses and definition of determinants. (Section 5.3 and start Section 5.4).
Lecture 23. Properties of determinants. (Section 5.4).
The Algebra problems are located at the end of each chapter of the Algebra Notes booklet. They
are also available from the course module on the UNSW Moodle server. The problems marked [R]
form a basic set of problems which you should try first. Problems marked [H] are harder and can
be left until you have done the problems marked [R] . You do need to make an attempt at the [H]
problems because problems of this type will occur on tests and in the exam. If you have difficulty
with the [H] problems, ask for help in your tutorial. Questions marked with a [V] have a video
solution available from the course page for this subject on Moodle. The problems marked [X] are
intended for students in MATH1141 – they relate to topics which are only covered in MATH1141.
Extra problem sheets for MATH1141 may be issued in lectures.
There are a number of questions marked [M], indicating that Maple is required in the solution
of the problem.
c
2019 School of Mathematics and Statistics, UNSW Sydney
1
Chapter 1
INTRODUCTION TO VECTORS
The aims of this chapter are to introduce the idea of “vector” and in a relatively informal and
intuitive manner, and to illustrate applications of these ideas to the geometry of lines and planes.
Until quite recently, the main applications of vectors had been in the physical and engineering
sciences. However the study of vectors has now become an important branch of modern pure
and applied mathematics, and vectors are now being used in such diverse fields as economics
and management science, psychology and the social sciences, chemistry and chemical engineering,
mechanical and electrical engineering, computer science, numerical analysis and computational
mathematics.
The definition of vectors used in mathematics courses is essentially algebraic in nature whereas
the one use by physicists is geometric. We shall begin with the geometric approach then we shall
introduce the algebraic definition and show how they relate to one another. As we shall see however,
the algebraic definition of a vector is not limited to describing quantities that arise in physics and
engineering.
writing an arrow above the letter (as in ~a). Because the properties of scalar and vector quantities
are quite different, it is vital that you distinguish them, especially in solutions to problems.
The magnitude of the vector a is usually denoted by |a|. Note that this is always a non-negative
real number, and that |a| = 0 only when a is the zero vector, usually denoted by 0.
Definition 1. The zero vector is the vector 0 of magnitude zero, and undefined
direction.
There are two equivalent ways to add two vectors together. The first addition rule is often
known as the triangle law for vector addition.
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.1. VECTOR QUANTITIES 3
c
b
c=a+b
a
Figure 2: Addition of Vectors.
The second addition rule is known as the parallelogram law for vector addition.
b c=a+b
a
Figure 3: Addition of Vectors.
These laws, which follow from basic geometry (see Figures 4 and 5), assert that it makes no
difference in what order, or in what grouping we add vectors.
b b
a+ a
b b+
a
c
2019 School of Mathematics and Statistics, UNSW Sydney
4 CHAPTER 1. INTRODUCTION TO VECTORS
c c
+c c)
b )
( b+
c
a+ a+
b+
(
b b b
a+
a a
Figure 5: Associative Law of Vector Addition.
The law of addition can easily be extended to cover the zero vector by the natural condition
that a + 0 = 0 + a = a for all vectors a. We then can introduce the negative of a vector and the
subtraction of vectors.
b + QP = a
−−→
b
QP = a − b
−−→ O Q
Hence the diagonal QP represents the difference b
a − b. Moreover, a − b is the vector which can be
represented by the arrow from the terminal point of b Figure 6: Subtraction of Vectors.
to the terminal point of a.
We now define the operation of multiplying a vector by a real number. Roughly speaking, to
multiply a vector a by a real number λ, all we do is stretch the vector by a factor of λ, whilst
keeping its direction unchanged. We need to be careful if λ is not positive.
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.1. VECTOR QUANTITIES 5
1. If λ > 0, then λa is the vector whose magnitude is λ|a| and whose direction is
the same as that of a.
2. If λ = 0, then λa = 0.
3. If λ < 0, then λa is the vector whose length is |λ||a| and whose direction is
the opposite of the direction of a.
−a = (−1)a.
Figure 7: Scalar Multiplication.
We have already seen the commutative and associative laws of vector addition. There are some
other important properties of scalar multiplication and vector addition. Let a and b be vectors, λ
and µ be real numbers, then:
The vector distributive law for the case that λ > 0 follows from properties of similar triangles and
Figure 8. The proof for the other cases and the proofs of the other two laws are left as exercises
for the 1141 students.
λb
a λa
Figure 8: Vector Distributive Law.
c
2019 School of Mathematics and Statistics, UNSW Sydney
6 CHAPTER 1. INTRODUCTION TO VECTORS
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.1. VECTOR QUANTITIES 7
−→ −−→
[X] Example 4. Let OA = a and OB = b. Prove that P is a point on the line AB between A
−−→
and B if and only if OP = (1 − λ)a + λb for some real number 0 < λ < 1.
Since 0 < λ < 1, the point P lies on AB between A and Figure 10.
B.
[X] Example 5. Prove that the three medians of a triangle are concurrent.
Proof. Name the vertices of a triangle by O, A and B. Let D, E, F be the midpoints of OB, OA, AB
respectively. Suppose that AD and BE intersect at G.
−→ −−→
Let OA = a and OB = b. Hence
−−→ 1 −−→ 1
OE = a and OD = b.
2 2
Since G lies on both AD and BE and inside the triangle,
B
from Example 4 there exist real numbers λ and µ such that
−−→ 1 1
OG = (1 − λ) a + λ b = (1 − µ) b + µ a .
2 2 D F
By rearranging terms, we get G
1 1 O A
(1 − λ) a − µ a = (1 − µ) b − λ b.
2 2 E
Since a cannot be a non-zero scalar multiple of b, we have
Figure 11.
1 1
(1 − λ) − µ = 0 and (1 − µ) − λ = 0.
2 2
By solving the above simultaneous equations, we have λ = 23
−−→
and µ = 32 . So we have OG = 13 (a + b).
−−→ −−→ −−→
Since F is the midpoint of AB, so OF = 21 (a + b). Thus OG and OF are in the same direction.
Hence G lies on OF and therefore the three medians of the triangle OAB are concurrent.
Many calculations with vector quantities in the plane can be done geometrically using scale
diagrams.
c
2019 School of Mathematics and Statistics, UNSW Sydney
8 CHAPTER 1. INTRODUCTION TO VECTORS
Example 6. A yacht sails from a pier in the direction N 60◦ E for 15 km, then turns to N 45◦ W
for 10 km. What are the distance and the bearing of the yacht from the pier?
p
Hence OQ ≈ (5.919)2 + (14.571)2 ≈ 15.73,
−1 14.571 30◦
θ ≈ tan ≈ 67◦ 54′ . E
5.919 O
Here, θ is angle of OQ measured from the east. The
distance and the bearing of the yacht from the pier are Figure 12.
then 15.73 km and N 22◦ 6′ E. ♦
−−→ −−→
In adding the two displacement vectors OP and P Q, we add and subtract displacements in the
−−→
east-west and the north-south direction. We then specified the sum OQ by its length and direction.
We shall see how this relates to the algebraic definition of vector in the next section.
Note. For many of the problems in physical world, the vector quantities are in three dimensions
rather than two. At least in theory, the same methods in this section could be used to solve such
problems. Each vector could be represented as an arrow in space, and then we can add two vectors
and multiply a vector by a scalar.
In Example 6 of Section 1.1.2, if we denote the vector of length 1 km towards the east by i and the
vector of length 1 km towards the north by j, by the definition of geometric vectors we can write
−−→
OP = 15 cos 30◦ i + 15 sin 30◦ j,
−−→
P Q = −10 cos 45◦ i + 10 sin 45◦ j, and
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.2. VECTOR QUANTITIES AND Rn 9
There is no reason why we cannot generalise this to all two dimensional vectors, three dimen-
sional vectors and beyond.
1.2.1 Vectors in R2
We first choose two vectors, conventionally denoted by i
and j, of unit length, and at right angles to each other so j
i
that j is pointing at an angle of π2 anticlockwise from i. The
vectors i and j are known as the standard basis vectors for
Figure 13.
R2 . See Figure 13.
As shown in Figure 14, every vector a can be ‘resolved’
(in a unique way) into the sum of a scalar multiple of i plus
a scalar multiple of j. That is there are unique real numbers
a
a1 and a2 such that a = a1 i + a2 j. If the direction θ of a is a2 j
measured from the direction of i, then these scalars can be
θ
easily found using the formulae
a1 i
a1 = |a| cos θ, and a2 = |a| sin θ. Figure 14.
c
2019 School of Mathematics and Statistics, UNSW Sydney
10 CHAPTER 1. INTRODUCTION TO VECTORS
λa1
2. the coordinate vector for λa is .
λa2
Proof. The basis vectors are i and j. We first give a detailed proof of the second part.
λa1
Hence the coordinate vector for λa is .
λa2
For the first part, by associative, commutative and distributive laws, we have
a + b = (a1 i + a2 j) + (b1 i + b2 j)
= (a1 i + b1 i) + (a2 j + b2 j)
= (a1 + b1 )i + (a2 + b2 )j
a1 + b1
Hence the coordinate vector for a + b is .
a2 + b2
We then can define the mathematics structure R2 , which is the set of 2-vectors, by
2 a1
R = : a1 , a2 ∈ R ,
a2
a1 b
with addition and multiplication by a scalar defined by — for any , 1 ∈ R2 and λ ∈ R,
a2 b2
a1 b1 a 1 + b1 a1 λa1
+ = and λ = .
a2 b2 a 2 + b2 a2 λa2
The elements in R2 are called vectors and sometimes column vectors. It is obvious that the set
R2 is closed under addition and scalar multiplication. Like geometric vectors, the vectors in R
2
0
also obey the commutative, associative, and distributive laws. There is a zero vector 0 = and
0
−a1 a1
a negative for any vector a = .
−a2 a2
1.2.2 Vectors in Rn
The concept of components can easily be generalised from two dimensions to any number of di-
mensions.
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.2. VECTOR QUANTITIES AND Rn 11
Note. We say that two vectors in Rn are equal if the corresponding components are equal. In
a1 b1
.. ..
other words . = . if and only if a1 = b1 , . . . , an = bn .
an bn
0.5
1 1.4
Example 1. 1. Clearly 0 ∈ R3 ; 4
2.3 ∈ R .
1
−4.1
1 2
2. 2 and 3 are different elements of R3 .
♦
3 1
a1 b1
a2 b2
Definition 2. Let a = . , b = . be vectors in Rn and λ be a real number.
.. ..
an bn
a 1 + b1
a 2 + b2
We define the sum of a and b by a + b = . .
..
a n + bn
λa1
λa2
We define the scalar multiplication of a by λ by λ a = . .
..
λan
c
2019 School of Mathematics and Statistics, UNSW Sydney
12 CHAPTER 1. INTRODUCTION TO VECTORS
Note. The addition rule tells us how to add two vectors with the same number of components.
The sum of vectors with different numbers of components is not defined.
4 2 4+2 6
Example 2. 1. −3 + 4 = −3 + 4 = 1 .
5 −2 5 + (−2) 3
1.3 3.2 4.5
2 −2 0
2.
0 + 2 = 2 .
1 1 2
1 3
3. 3 2 = 6 .
3 9
−1 −π
2 2π
4. π
0 = 0 .
5 5π
2
1 0
5. 2 +
4 is not defined! ♦
3
−1
[X] Proof. We prove the commutative law, while proving the associative law will be left as an
exercise.
u1 v1
.. ..
Let u = . and v = . .
un vn
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.2. VECTOR QUANTITIES AND Rn 13
u1 v1
.. ..
u+v = . + .
un vn
u1 + v 1
..
= . (definition of addition in Rn )
un + v n
v 1 + u1
..
= . (commutative law of real numbers)
v n + un
v1 u1
.. ..
= . + . (definition of addition in Rn )
vn un
= v+u
1. a + 0 = 0 + a = a.
a1 b1 a 1 − b1
2. ... − ... = ... .
an bn a n − bn
3. a + (−a) = (−a) + a = 0.
0
0 0
Example 3. 1. 0 is the zero vector in R3 , while
0 is the zero vector in R
4
0
0
c
2019 School of Mathematics and Statistics, UNSW Sydney
14 CHAPTER 1. INTRODUCTION TO VECTORS
4 2 4−2 2
2. −3 − 4 = −3 − 4 = −7 .
5 −2 5 − (−2) 7
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.3. Rn AND ANALYTIC GEOMETRY 15
(iii) specify three directions, usually called the x, y and z directions, each at right angles to the
others.
We choose unit vectors i, j, k, respectively, in the x-direction, y-direction, z-direction. The coor-
dinates of a point A with respect to the basis {i, j, k} are given by a 3-tuple (x1 , y1 , z1 ) such that
c
2019 School of Mathematics and Statistics, UNSW Sydney
16 CHAPTER 1. INTRODUCTION TO VECTORS
A is x1 units from the origin in the x-direction, y1 units from the origin in the y-direction, and z1
−→
units from the origin in the z-direction.
The position vector of A, OA, is a = x1 i + y1 j + z1 k,
x1
which can also be represented by y1 ∈ R3 .
z1
z-direction
k-direction
z-axis
z1
b
A A
a z1 k
y1
O O
y-direction x1 i j-direction
x1 y-axis
y1 j
i-direction
x-direction x-axis
Figure 18: Right-handed Coordinate System.
The coordinate system in Figure 18 is what is called right-handed Cartesian coordinate
system. Using your right hand, if you point your index finger in the x-direction, your middle finger
in the y-direction, then your extended thumb will point in the z-direction. This is the orientation
that is conventionally used in mathematics and physics.
Again once we have set up our coordinate system, each point in space corresponds uniquely to
a 3-tuple and uniquely to a 3-vector (and vice-versa). It is rather hard to draw pictures of objects
in three dimensions (on two dimensional paper!). Nevertheless, this identification will provide us
with an important way of visualising relationships between vectors in R3 .
Since the basis vectors chosen arepof unit length and are perpendicular to each other, by Pythago-
−→
ras’
theorem
the length of OA is x21 + y12 + z12 . Let B be another point with position vector
x2
−−→
b = y2 , the distance between A and B is the same as the length of the vector b − a = AB, i.e.
z2
p
dist(A, B) = (x1 − x2 )2 + (y1 − y2 )2 + (z1 − z2 )2 .
−→
Similar to R2 , if a = OA, the length of a is dist(O, A). Two non-zero vectors a and b in R3 are
parallel if and only if there is a non-zero real number λ such that a = λb.
1.3.3 n-dimensions
We have used elements in R2 and R3 to represent the position vectors of points in the plane and in
the three-dimensional space, respectively. How do we interpret n-vectors in Rn as position vectors
of points in an n-dimensional space which generalises the 2-dimensional and the 3-dimensional
spaces? To study the geometry in the n-dimensional space, we need the notions of length and
direction.
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.3. Rn AND ANALYTIC GEOMETRY 17
In R2 or R3 , two non-zero vectors a and b are parallel if and only if there exists a non-zero real
number λ such that b = λa. We use this idea as definition.
4 8 8 4
3 6
0 1
The vectors
2 and 4 are not parallel because we cannot find a scalar λ such that
1 2
6 3
1 0
= λ . ♦
4 2
2 1
Definition 2. Let ej be the vector in Rn with a 1 in the jth component and 0 for
all the other components. Then the n vectors e1 , e2 , . . . , en are called the standard
basis vectors for Rn .
We imagine that we can choose a point in the n-dimensional space as the origin O and n axes in
the directions of these n standard basis vectors respectively. Then we can define the coordinates
of a point A with respect to this coordinate system in the same way as in 3-dimensional space. For
instance, if A is ai units from O in the direction of ei for each 1 6 i 6 n, the coordinates of A
are given by an n-tuple
(a1 , . . . , an ). Thus the coordinate vector, or the position vector, of the
a1
..
point A is a = . ∈ Rn . We can also discuss n-dimensional geometric vectors. For example,
an
−→
the vector OA, which has the initial point O and the terminal point A, represents a.
c
2019 School of Mathematics and Statistics, UNSW Sydney
18 CHAPTER 1. INTRODUCTION TO VECTORS
Example 3. i) The coordinate vector for the displacement from the point A, with coordinates
(2, 3) to the point B with coordinates (−1, 2) is
−−→ −−→ −→ −1 2 −3
AB = OB − OA = − = .
2 3 −1
ii) The coordinate vector for the line segment joining the point A with coordinates (2, 4, −5, 3)
to the point B with coordinates (−3, 4, 0, 4) is
−3 2 −5
−−→ 4
4 0
AB =
0 − −5 = 5 .
4 3 1
−1
iii) If the position vector of the point A is 2 and the displacement from A to B has
2
4
coordinate vector 2 , then the position vector for B is
−3
−1 4 3
−−→ −→ −−→
OB = OA + AB = 2 + 2 = 4 .
2 −3 −1
We can also define the length of a vector and the distance between two points which have
coordinate vectors in Rn .
a1
Definition 3. The length of a vector a = ... ∈ Rn is defined by
an
q
|a| = a21 + · · · + a2n .
1
−3
2
Example 4. Find the lengths of a = and b =
2.
−4 −6
4
√ √ √ √
Solution. |a| = 4 + 16 = 2 5; |b| = 1 + 9 + 4 + 36 + 16 = 66. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.4. LINES 19
Definition 4. The distance between two points A and B with position vectors
−−→
in Rn is the length of the vector AB.
a1 b1
.. ..
Then, if a = . is the coordinate vector of A and b = . is the coordinate vector of B,
an bn
we have
b1 − a 1
−−→
AB = b − a = ... ,
bn − a n
Example 5. Find the distance between the point A with coordinates (3, −2, 5, 1) and the point B
with coordinates (−1, 2, −8, 4).
−4
−−→ 4 −−→ √ √
Solution. As AB =
−13 , the distance is AB = 16 + 16 + 169 + 9 = 210. ♦
3
Note. Until now, we distinguish an n-dimensional space from Rn which contains all the position
vectors of the points in the space. From now on, we shall blur the difference. So we often refer
to R2 as the plane and the R3 as the three dimensional space. In our algebra notes, a point in an
n-space is usually referred as an n-tuple
of coordinates (a1 , . . . , an ) and the corresponding position
a1
..
vector is denoted by an n-vector . .
an
1.4 Lines
Suppose now that we wish to express a line in the plane in terms of coordinate vectors. One form
for the equation of a line is y = mx + b, where m is the slope of the line and b is the y-intercept.
Since a line is really just a set of points, each point on the line will have a corresponding
coordinate vector in R2 . There is a simple way to describe the coordinate vector of each point on
a line in the plane.
c
2019 School of Mathematics and Statistics, UNSW Sydney
20 CHAPTER 1. INTRODUCTION TO VECTORS
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.4. LINES 21
3
5
Example 2. Write a parametric vector form for the line in R4 spanned by the vector
−6 .
8
3
5
A solution is x = λ
−6 for λ ∈ R. ♦
8
Note. A given line does not have a unique parametric vector form. The following are also
parametric vector forms of the same line.
−3 9
−5 15
x = λ
6
and x = λ
−18 .
−8 24
On the other hand, the name of the parameter in the expression is not important. For example
the forms
3 3
5 5
x = µ
−6 for µ ∈ R and x = s −6 for s ∈ R
8 8
c
2019 School of Mathematics and Statistics, UNSW Sydney
22 CHAPTER 1. INTRODUCTION TO VECTORS
Example 3. Find aparametric vector form of the line through the point (2, −3, 1, 6) parallel to
1
4
the vector v =
−6 .
2
Solution. Each point on the line has coordinate vector of the form
2 1
−3 4
x=
1 + λ −6 , for some λ ∈ R. (*)
6 2
♦
Note. Once again, a line may have different parametric vector forms. For instance, the following
line
3 1
1
x= + λ 4 , for some λ ∈ R. (**)
−5 −6
8 2
2 1 3
−3 4 1
is one which passes through (3, 1, −5, 8) parallel to v. Note that
1 + −6 = −5 , so
6 2 8
(3, 1, −5, 8) is another point on the line with equation (*). Thus both (*) and (**) are parametric
vector forms of the same line.
1.4.1 Lines in R2
We can write the equation of a line in Cartesian form y = mx + d and also in parametric vector
form x = a + λv. It is often necessary to convert between these two forms.
Conversion from Cartesian form to parametric vector form.
The equation y = mx + d can be converted to parametric vector form as follows:
x
Suppose x = is the position vector of a point on the line and set x = λ. Then, since
y
y = mλ + d, we obtain
λ 0 λ 0 1
x= = + = +λ .
mλ + d d mλ d m
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.4. LINES 23
The equation y = mx + d is a special case of the more general linear equation in two variables
x and y, which is given by
ax + by = c,
where a, b, and c are fixed numbers. If b 6= 0, this linear equation can easily be converted to the
y = mx + d form by dividing by b. Alternatively, the equation ax + by = c can be converted
directly to parametric vector form by setting either x or y as a parameter. The following example
illustrates this technique.
ii) Note that 2x = 8 means x = 4. So the x has a fixed value but the y value varies. We need to
set y = λ. Hence a parametric vector form for the line is
x 4 4 0
x= = = +λ , for λ ∈ R.
y λ 0 1
iii) In this case the equation fixes y = 2, whereas x can have any real value. We therefore set x = λ
as the parameter and obtain
x λ 0 1
x= = = +λ , for λ ∈ R.
y 2 2 0
c
2019 School of Mathematics and Statistics, UNSW Sydney
24 CHAPTER 1. INTRODUCTION TO VECTORS
1.4.2 Lines in R3
A line in R3 is still of the form x = a + λv, but now the vectors x, a and v each have three
components or three coordinates.
An alternative Cartesian (or symmetric) form for the equation of a line is sometimes used in
engineering. This form can be obtained as follows:
x a1 v1
Let x = y , a = a2 and v = v2 be vectors in R3 . Then the parametric vector
z a3 v3
v1
equation of the line through (a1 , a2 , a3 ) parallel to v2 is
v3
x a1 v1
x = y = a2 + λ v 2 .
z a3 v3
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.4. LINES 25
Eliminating the parameter λ yields (if all vi 6= 0), the Cartesian form
x − a1 y − a2 z − a3
= = (= λ).
v1 v2 v3
If v1 , v2 or v3 is 0, then x, y or z will, respectively, be constant.
c
2019 School of Mathematics and Statistics, UNSW Sydney
26 CHAPTER 1. INTRODUCTION TO VECTORS
x−4 2y + 3 z
= = − = λ.
7 2 6
By rearranging terms, we get the parametric equations
3
x = 7λ + 4, y =λ− , z = −6λ.
2
In vector form, we have
x 7λ + 4 4 7
y = λ − 3 = −3 + λ 1 .
2 2
z −6λ 0 −6
Then we shall get the same parametric vector form as the one we have obtained by Method 1. ♦
x − 2y = 4 and z = 1.
Example 10. Find a parametric vector form of the line through the points A (1, 3, −4) and
B (2, 0, 6).
Solution. The line passes through the point A (1, 3, −4) and it is parallel to the vector
2 1 1
−−→
AB = 0 − 3 = −3 .
6 −4 10
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.4. LINES 27
Example 11. Find the equation of the line in R4 through the points A (1, 0, 3, −2) and B (2, 1, 0, −1)
in parametric vector form and Cartesian form.
2 1 1
1 0 1
Solution. The line is parallel to the vector
0 − 3 = −3 . Thus the line in
−1 −2 1
parametric vector form is
1 1
0 1
x=
3 + λ −3 , for λ ∈ R.
−2 1
The Cartesian form of the line is
x1 − 1 x2 x3 − 3 x4 + 2
= = = .
1 1 −3 1
♦
More generally, a vector equation of the line joining A to B, with position vectors a and b
respectively, is
x = a + λ(b − a) = (1 − λ)a + λb, λ ∈ R.
Different values of λ correspond to different points
on this line. For example, the value λ = 51 , gives the
position vector B
b
1
4 1 1 λ= 5
x= a + b = a + (b − a). P b
5 5 5 A b
b
x
This vector can be written as
a
−→ 1 −−→ O
x = OA + AB.
5 Figure 22.
In R2 or R3 , it is the position vector of the point P
which divides AB in the ratio 1 : 4 (see Figure 22).
c
2019 School of Mathematics and Statistics, UNSW Sydney
28 CHAPTER 1. INTRODUCTION TO VECTORS
(1 − λ)a + λb,
Example 12. Find the medians of the triangle with vertices A (1, 2, 3), B (−2, 1, −4) and C (4, −1, 2).
Solution. A median of a triangle is a line through a vertex and the midpoint of the opposite side.
The median through A is therefore the line which passes through the point A and the midpoint P
of the line segment BC, as shown in Figure 23.
Let a, b, c be the position vectors of A, B, C, respectively.
Since P is the midpoint of BC, the coordinate vec-
tor of P is
−2 4 1
1 1
(b + c) = 1 + −1 = 0 .
2 2 A
−4 2 −1
and hence B
1 1 0
−→
AP = 0 − 2 = −2 . P
−1 3 −4 C
−→
Since A lies on the median and the vector AP is par-
allel to the median, we can write
O
1 0
x = 2 + λ1 −2 , λ1 ∈ R
Figure 23.
3 −4
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.5. PLANES 29
3 −2 4 3
Example 13. Are the lines x = −1 + λ1 −3 and x = 0 + λ 2 29 parallel?
2 4 1 −6
Solution. Since two vectors
are parallel if and only if one is a non-zero scalar multiple of the
3 −2
3
other, and 29 = − −3 , we can conclude that the two lines are parallel. ♦
2
−6 4
1.5 Planes
We have seen that the set of all vectors x ∈ Rn which are scalar multiples of a fixed non-zero vector
v ∈ Rn , i.e. the set {x ∈ Rn : x = λv, for λ ∈ R}, has a simple geometric interpretation as a
line through the origin. In this section we shall show that the set of all sums of scalar multiples of
two non-zero non-parallel vectors v1 and v2 has a geometric interpretation as a plane through the
origin. Furthermore, we shall look at how to find equations for planes.
−3 1 3
Example 1. i) The vector 2 is a linear combination of the vectors 0 and −1 since
6 2 0
−3 1 3
2 = 3 0 + (−2) −1 .
6 2 0
0 1 3
ii) The vector 1 is not a linear combination of 0 and −1 since there are no real
2 2 0
numbers λ1 and λ2 such that
0 1 3
1 = λ1 0 + λ2 −1 .
2 2 0
c
2019 School of Mathematics and Statistics, UNSW Sydney
30 CHAPTER 1. INTRODUCTION TO VECTORS
Definition 2. The span of two vectors v1 and v2 , written span(v1 , v2 ), is the set
of all linear combinations of v1 and v2 . That is, it is the set
span(v1 , v2 ) = {x : x = λ1 v1 + λ2 v2 , for some λ1 , λ2 ∈ R }.
Example 2. Given that v1 and v2 are non-zero, non-parallel vectors in R3 . What does span(v1 , v2 )
look like?
Take x in span(v1 , v2 ) . So x = λ1 v1 + λ2 v2 ,
for some λ1 , λ2 ∈ R. By the parallelogram law λ1 v 1 + λ2 v 2
of vector addition, x lies in the plane passing λ2 v 2
through O and contains v1 and v2 . v2
x
Conversely, suppose that X is a point with
position vector x on this plane and OX is nei-
ther parallel to v1 nor parallel to v2 . Since v1
O λ1 v 1
v1 and v2 are non-zero and non-parallel, there
exists a parallelogram OAXB, such that OA
and OB are parallel to v1 and v2 respectively. B X
−→
Note also that, in this case, OA = λ1 v1 and v2
−−→ x
OB = λ2 v2 , for some real numbers λ1 , λ2 . By
the parallelogram law, we have x = λ1 v1 + λ2 v2
which is in span(v1 , v2 ).
When OX is parallel to v1 , we have x = λ1 v1 O v1 A
which is also in the span. Similarly, x is also in Figure 24.
the span when OX is parallel to v2 .
Hence, span(v1 , v2 ) is the plane through the origin and parallel to v1 and v2 with equation
x = λ1 v 1 + λ2 v 2 , for some λ1 , λ2 ∈ R.
♦
Note that The above construction does not work if v1 and v2 are parallel or one of them is 0.
Theorem 1. In R3 , the span of any two non-zero non-parallel vectors is a plane through the origin.
We can extend this to Rn using:
Definition 3. A plane through the origin is the span of any two (non-zero)
non-parallel vectors.
3 4
Example 3. Describe geometrically span −4 , 2 , and decide if it is a plane or a line.
5 7
Solution. By definition, the span is the set of all vectors x ∈ R3 such that
3 4
x = λ1 −4 + λ2 2 for some λ1 , λ2 ∈ R.
5 7
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.5. PLANES 31
3 4
As −4 is not a scalar multiple of 2 , the span is a plane through the origin parallel to the
5 7
3 4
vectors −4 and 2 . ♦
5 7
3 6
5 10
Example 4. Describe geometrically span
2 , 4 and decide whether it is a plane or a
7 14
8 16
line.
3 6
5 10
x = λ1 2
+ λ2 4
for some λ1 , λ2 ∈ R.
7 14
8 16
6 3 3
10 5 5
As
4 = 2 2 , the span is the line x = λ 2 for some λ ∈ R. ♦
14 7 7
16 8 8
The span of two non-zero non-parallel vectors is a plane through the origin. For planes that do not
pass through the origin we use a similar approach to that we use for lines.
c
2019 School of Mathematics and Statistics, UNSW Sydney
32 CHAPTER 1. INTRODUCTION TO VECTORS
S = {x ∈ R3 : x = a + λ1 v1 + λ2 v2 for λ1 , λ2 ∈ R}. S
This set consists of all the vectors in R3 for which
x − a ∈ span(v1 , v2 ). That is, S contains all the x−a
points x for which x − a lies in the plane through
the origin spanned by v1 and v2 . Thus S is the plane
in R3 which passes through the point with position x
vector a, and is parallel to v1 and v2 . The picture a
is shown in Figure 25.
O λ2 v 2
So in R3 ,
λ1 v 1
x = a + λ1 v 1 + λ2 v 2 , for λ1 , λ2 ∈ R
Figure 25: x = a + λ1 v1 + λ2 v2 .
is an equation of the plane passing through a and
parallel to the non-zero non-parallel vectors v1 , v2 .
This is called a parametric vector form of the plane.
As before we shall use this as a way of defining what we mean by a plane in Rn .
Note. For a given plane, there is not a unique parametric vector form. For example, a plane
parallel to v1 and v2 is also parallel to v1 + v2 and v1 − v2 .
From the definition, a point P with position vector p ∈ Rn is said to lie on the plane through
a parallel to v1 , v2 , if there exist real numbers λ1 , λ2 such that p = a + λ1 v1 + λ2 v2 .
Example 5. Find
a parametric
vector form for the plane through the point (1, 3, 0), parallel to
4 −2
the vectors 2 and 3 .
9 4
Solution. One answer (there are many possible!) is
1 4 −2
x = 3 + λ1 2 + λ2 3 , for λ1 , λ2 ∈ R.
0 9 4
♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.5. PLANES 33
Example 7. Let A, B and C be three points in Rn whose coordinate vectors with respect to some
coordinate system are a, b and c respectively. Find a parametric vector form for the plane which
passes through A, B and C.
−−→ −→
Solution. Consider the plane through A parallel to AB and AC. This has a parametric vector
form:
−−→ −→
x = a + λ1 AB + λ2 AC = a + λ1 (b − a) + λ2 (c − a), for λ1 , λ2 ∈ R.
Example 8. Find a parametric vector form for the plane through the three points A (2, −1, 3), B
(−1, 4, 4) and C (3, −1, 2).
−1 2 −3 4
−−→ −−→
Solution. The plane is parallel to AB = 4 − −1 = 5 and BC = −5 , and
4 3 1 −2
passes though the point A (2, −1, 3). Hence a parametric vector form for the plane is
2 −3 4
x = −1 + λ1 5 + λ2 −5 , for λ1 , λ2 ∈ R.
3 1 −2
c
2019 School of Mathematics and Statistics, UNSW Sydney
34 CHAPTER 1. INTRODUCTION TO VECTORS
represents plane when not all a, b, c are zero. In Rn , n > 3, if not all a1 , a2 , . . . , an are zero, an
equation of the form
a1 x1 + a2 x2 + · · · + an xn = d
is called a hyperplane.
Example 9. Find a parametric vector form for the plane x1 − 3x2 + 4x3 = 4, and hence find a
point on the plane and two vectors parallel to the plane.
Solution. We set x2 = λ1 and x3 = λ2 . Then solving for x1 gives x1 = 4 + 3λ1 − 4λ2 . Hence
x1 4 + 3λ1 − 4λ2 4 3 −4
x = x2 = λ1 = 0 + λ1 1 + λ2 0 ,
x3 λ2 0 0 1
3
for λ1 , λ2 ∈ R. Thus the plane passes through the point (4, 0, 0) and is parallel to 1 and
0
−4
0 . ♦
1
As we did for lines, we can find a Cartesian equation for a plane by eliminating the parameters
λ1 , λ2 from the parametric vector form.
1 1 −1
Example 10. Find the Cartesian form of the plane x = 1 + λ 1 + µ 0 for λ, µ ∈ R.
2 1 2
x1
Solution. Let x = x2 . By comparing the components, we get the following parametric
x3
equations
x1 = 1 + λ − µ, x2 = 1 + λ and x3 = 2 + λ + 2µ.
Hence from the second equation, we get λ = x2 − 1. Substituting this into the first equation, we
have x1 = 1 + (x2 − 1) − µ, so µ = x2 − x1 . Now substitute these values of λ and µ into the third
parametric equation. The Cartesian form of the plane is, therefore,
c
2019 School of Mathematics and Statistics, UNSW Sydney
1.6. VECTORS AND MAPLE 35
c
2019 School of Mathematics and Statistics, UNSW Sydney
36 PROBLEMS FOR CHAPTER 1
1. [R][V] Given that ABC, DEF, and OGH are equally spaced parallel lines, as are ADO, BEG
and CF H. P is the mid point of AD.
A B C
P b
D E F
O G H
−−→ −→
If OH = h and OA = a, express the following in terms of a and h.
−−→ −−→ −−→ −−→ −−→
a) OC, b) HA, c) GC, d) OP , e) GP .
2. [R] Simplify
−−→ −−→ −→ −−→ −−→ −−→ −−→
a) AB − OB + OA, b) AB − CB + 3DA + 3CD.
a) If M is the midpoint of the line segment AB and P is the midpoint of the line segment
−−→ −−→
CB express the vectors OM and OP in terms of a, b, and c.
−−→ −→
b) Show that M P is parallel to AC and has half its length.
5. [H][V] Given a convex quadrilateral ABCD, prove, using vectors, that the quadrilateral formed
by joining the midpoints of AB, BC, CD, and DA is a parallelogram.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 1. INTRODUCTION TO VECTORS 37
6. [R] Use geometric vectors to solve the following problems. In each case, draw a careful picture
and then use trigonometry to find the answer. If your picture is accurate, you may wish
to use a ruler and protractor to confirm your result.
a) An ant crawls 10 cm due east in a straight line and then crawls 5 cm northeast in a
straight line. What is the ant’s final displacement from its starting point?
b) An ant is standing at the western edge of a moving walkway which is moving at 12
cm per sec in the direction due South. The ant starts to walk at 5 cm/sec across the
walkway in the direction perpendicular to its edge. If the walkway is 40 cm wide, find
the displacement of the ant from its starting point just as it steps off the walkway.
c) An observer on a wharf sees a yacht sailing at 15 km per hour southeast. A sailor on
the yacht is watching a container ship and sees it sailing at 25 km per hour due north.
What is the velocity of the container ship as seen by the observer on the wharf?
d) A rower is rowing across a river. His rowing speed is 2 km per hour and there is a
current flowing in the river at 1 km per hour. Find the direction that the rower must
row to go directly across the river. If the river is 300 metres wide, how long will it
take him to cross the river?
8. [X] Let O, A, B, C be points in a plane. Suppose that X is the midpoint of BC, Y is the
point on AC with AY : Y C = 3 : 1, and Z is the point on AB with AZ : ZB = 3 : 1. Let
−→ −−→ −−→
OA = a, OB = b, and OC = c.
−−→ −−→ −→
a) Write down the vectors OX, OY and OZ in terms of a, b, c.
−→
b) Let T be the point on AX with AT : T X = 6 : 1. Write OT in terms of a, b, c.
c) Show that T lies on both CZ and Y B.
c
2019 School of Mathematics and Statistics, UNSW Sydney
38 PROBLEMS FOR CHAPTER 1
−1 10
0
d) u = 3 ,v= 2 ,w= ;
0
5 −3
e) u = 2i + 3j − 2k, v = i − 2j + k, w = −i + j − k.
10. [R] A car travels 3km due North then 5km Northeast. Use coordinate vectors to find the
distance and direction from the starting point.
11. [R] Solve Problems 6 (a) and (b) using coordinate vectors.
a1 a2
12. [R] Suppose that v = b1 and w = b2 are vectors in R3 ; λ and µ are real numbers.
c1 c2
Prove the scalar distributive law (λ + µ)v = λv + µv and the vector distributive law
λ(v + w) = λv + λw.
13. [X] Prove the associative law of vector addition in Rn . (Proposition 2 on page 12).
a) A = (1, 2, 3), B = (−2, 3, 4), C = (−3, −4, 7), D = (4, −6, −9);
b) A = (3, 2, 5), B = (5, −3, −6), C = (−2, 3, 7), D = (0, −2, −4);
c) A = (12, −4, 6), B = (2, 6, −4), C = (5, −2, 9), D = (0, 3, 4).
Do any of these sets of 4 points form a parallelogram?
17. [R] Prove that A(1, 2, 1), B(4, 7, 8), C(6, 4, 12) and D(3, −1, 5) are the vertices of a parallel-
ogram. Draw and label the parallelogram.
18. [R] Show that the points A(1, 2, 3), B(3, 8, 1), C(7, 20, −3) are collinear.
19. [R][V] Show that the points A(−1, 2, 1), B(4, 6, 3), C(−1, 2, −1) are not collinear.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 1. INTRODUCTION TO VECTORS 39
21. [H] If A(−1, 3, 4), B(4, 6, 3), C(−1, 2, 1) and D are the vertices of a parallelogram, find all the
possible coordinates for the point D.
22. [H] Consider three non-collinear points D, E, F in R3 with coordinate vectors d, e and f .
There are exactly 3 points in R3 which, taken one at a time with D, E and F, form a
parallelogram. Calculate vector expressions for the three points.
23. [R][V] Let A = (2, 3, −1) and B = (4, −5, 7). Find the midpoint of A and B. Find the point Q
on the line through A and B such that B lies between A and Q and BQ is three times as
long as AB.
24. [R] The coordinate vectors, relative to the origin O, of the points A and B are respectively a
and b. State, in terms of a and b, the position vector of the point T which lies on AB
−→ −→
and is such that AT = 2T B.
26. [R] For each of the following vectors, find its length and find a vector of length one (“unit”
vectors) parallel to it.
4
2
4 0
1
a = −4 , b=
0 , c=
1 .
2 −2
3
0
27. [R][V] Find the distances between each of the following pairs of points with coordinate vectors:
3 −2
8 −6 1 5 0 6
a) −4 , 1 ; b) 1 , −7 ; c)
1 ,
1 .
2 0 1 −7
4 3
4 7 6
28. [R] A triangle has vertices A, B and C which have coordinate vectors 1 , −4 and 2
7 6 8
respectively. Find the lengths of the sides of the triangle and deduce that the triangle is
right-angled.
c
2019 School of Mathematics and Statistics, UNSW Sydney
40 PROBLEMS FOR CHAPTER 1
33. [R] Find parametric vector forms for the following lines in R2 :
a) y = 3x + 4; b) 3x + 2y = 6; c) y = −7x;
d) y = 4; e) x = −2.
In each case indicate the direction of the line and a point through which the line passes.
34. [R] Find a parametric vector form and a Cartesian form for each of the following lines
2
Let Q be the point on AB such that AQ = AB.
3
a) Find q, the position vector of Q.
b) Find the parametric vector equation of the line that passes through P and Q.
36. [R] Decide whether each of the following statements is true or false.
2 4
a) The lines y = 3x − 4 and x = +λ are parallel.
1 12
3 6
b) The lines x = +λ and 2x + 3y = 8 are parallel.
−1 4
4 10
x + 10 z+3
c) The lines x = −1 + λ 2 and =y−7= are parallel.
5 4
2 8
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 1. INTRODUCTION TO VECTORS 41
3 10
d) The line x = −2 + λ 0 and the line
7 −4
x + 10 z+3
= and y = −5
5 −2
are parallel.
37. [X] Suppose A and B are points with coordinate vectors a and b, respectively. Write down a
parametric vector form for
38. [X] Give a geometric interpretation of the following sets. In each set, λ ∈ R.
1 −3
a) S = x : x = 3 + λ 1 for 0 6 λ 6 1 .
6 7
1 −2
2 5
b) S = x : x = 4 + λ 9 for − 1 6 λ 6 5 .
0 3
−7 6
6 0
−2 4
7 8
c) S = x : x = λ + (1 − λ) 3 for 0 6 λ 6 1.
2
−1 −5
5 4
1 3
4 0
d) S = x : x = + λ for λ > 0 .
−6 −1
2 5
3 6
e) S = x : x = 1 + λ −2 for |λ| > 2 .
−4 7
39. [R] Find a parametric vector form for the planes passing through the points
a) (0, 0, 0), (3, −1, 2), (1, 4, −6); b) (1, 4, −2), (2, 6, 4), (1, −10, 3).
c
2019 School of Mathematics and Statistics, UNSW Sydney
42 PROBLEMS FOR CHAPTER 1
40. [R] For each of the following sets of vectors, decide if the set is a line or a plane, give a point on
the line or plane, and give vectors parallel to the line or plane, i.e., geometrically describe
the sets.
1 −2
a) S = x : x = λ1 2 + λ2 3 for λ1 , λ2 ∈ R .
3 4
3 −2 4
1 1 −2
b) S = x : x = + λ1 + λ 2
for λ 1 , λ 2 ∈ R .
2 3 −6
4 2 −4
3 −9
2 −6
c) span
1 , −3 .
2 −6
1 4 8
d) S = x : x = 2 + y for y ∈ span
−1 , 2 .
3 2 4
x1 − 5 x2 + 6 x3 − 2 x4 + 1
= = = .
7 2 −3 −5
42. [R] Find parametric vector forms to describe the following planes in R3 .
a) x1 + x2 + x3 = 0. b) 3x1 − x2 + 4x3 = 12.
c) x2 + 6x3 = −1. d) x3 = 2.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 1. INTRODUCTION TO VECTORS 43
2
43. [H] Show that the line x = t 1
3
x−3 y+2
46. [H] Consider the line = = z − 1 and the plane 2x + y + 3z = 23 in R3 .
−2 3
a) Find a parametric vector form for the line.
b) Hence find where the line meets the plane.
48. [X] The following sets of points represent simple geometric figures in a plane. λ1 and λ2 are
real numbers. For each problem draw a sketch in the (λ1 , λ2 ) plane and a second sketch
in R2 , R3 or R4 (!!) as appropriate. For each problem identify the geometric shape.
0 1 2
a) S = x : x = + λ1 + λ2 for 0 6 λ1 6 1, 0 6 λ2 6 1 .
1 2 3
0 1 2
b) S = x : x = + λ1 + λ2 for 0 6 λ1 6 1, 0 6 λ2 6 λ1 .
1 2 3
2 4
c) S = x : x = λ1 1 + λ2 −2 for 0 6 λ1 6 6, 0 6 λ2 6 8 .
−2 3
2 4
d) S = x : x = λ1 1 + λ2 −2 for 0 6 λ1 6 6, 0 6 λ2 6 λ1 .
−2 3
c
2019 School of Mathematics and Statistics, UNSW Sydney
44 PROBLEMS FOR CHAPTER 1
2 4
1
e) S = x : x = λ1 + λ2 −2 for 0 6 λ1 6 6, 0 6 λ1 6 λ2 .
−2 3
2 −1
49. [X] Write down the sets of points corresponding to the following:
a) A “parallelogram” with the three vertices A(1, 3, 4, 2), B(−2, 1, 0, 5) and C(−4, 0, 6, 8).
Hint: Look at Question 48 a), and assume B and C are adjacent to A.
b) The triangle with the three vertices given in part a) of this question.
Hint: Look at Question 48 b).
c) All three parallelograms which have the three vertices given in part a).
x = a + s1 u1 + s2 u2 and x = b + t1 v 1 + t2 v 2 ,
x = s 1 e1 + s 2 e2 and x = e4 + t 1 e2 + t 2 e3
c
2019 School of Mathematics and Statistics, UNSW Sydney
45
Chapter 2
VECTOR GEOMETRY
“Why,” said the Gryphon, “you first form into a line along the sea-shore—”
“Two lines!’ cried the Mock Turtle.
Lewis Carroll, Alice in Wonderland.
In Chapters 1 and 4 we have shown how vectors can be used to solve geometric problems
involving points, lines and planes.
Our aim in this chapter is to show how vectors can be used to solve geometric problems involving
lengths, distances, areas, angles and volumes. For simplicity, and because of the fundamental
importance of two and three dimensions in the physical sciences and engineering, we will concentrate
on problems in two and three dimensions. However, we shall see that many of the two and three
dimensional results can be easily generalised to Rn . The key idea is to use theorems in R2 and
R3 to motivate definitions in Rn for n > 3.
2.1 Lengths
We have defined lengths of vectors in Rn and distance between two points on page 18. We expect
that lengths and distances in Rn should have the same essential properties as those in two and
three dimensional spaces. For example, we expect them to be real non-negative numbers. Some of
these essential properties for Rn are proved in the following proposition.
Proposition 1. For all a ∈ Rn and λ ∈ R,
1. |a| is a real number,
2. |a| > 0,
3. |a| = 0 if and only if a = 0,
4. |λa| = |λ| |a|.
Proof. From Definition 3 on page 18
q
|a| = a21 + · · · + a2n .
As a ∈ Rn , all a2k are real and non-negative, and hence a21 + · · · + a2n > 0. Thus, properties 1 and 2
hold.
c
2019 School of Mathematics and Statistics, UNSW Sydney
46 CHAPTER 2. VECTOR GEOMETRY
Property 3 holds since a sum of non-negative numbers is zero if and only if every term in the
sum is zero.
The proof of Property 4 is as follows. From the definitions of λa and the length of a vector, we
have p q
|λa| = (λa1 ) + · · · + (λan ) = |λ| a21 + · · · + a2n = |λ| |a|.
2 2
a1 b1
If we write a = a2 and b = b2 and use the formula for the length of c = a − b, we have
a3 b3
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.2. THE DOT PRODUCT 47
Notice that we have already met the expression for the dot product (see Example 4 of Section 5.2)
as the “scalar product” aT b, i.e., a · b = aT b.
For the special case of R2 or R3 , we then have
Students of physics and engineering should note that it is this geometric result which is usually
taken as the definition of the dot product in physics and engineering courses.
2 −1
Example 1. Find the dot product of 1 and 3 , and hence find the cosine of the angle
4 2
between the vectors.
Solution. The dot product is
2 −1
1 · 3 = −2 + 3 + 8 = 9,
4 2
√ √
and the lengths are 21 and 14, and hence
9 9
cos θ = p = √ .
(21)(14) 7 6
♦
Notice that the value of cos θ does not uniquely define the value of the angle θ. It is conventional
to define the angle between two vectors as an angle θ in the interval [0, π], so that the value of cos θ
does uniquely define the value of θ.
2
Example 2. On choosing the angle in the interval [0, π], the angle between the vectors 1 and
4
−1
3 of Example 1 is
2
9
θ = cos−1 √ = 1.018 . . . radians.
7 6
♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
48 CHAPTER 2. VECTOR GEOMETRY
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.2. THE DOT PRODUCT 49
−1 2
2 −4
Example 3. Find the angle between a =
−3 and b = 0 .
−1 −1
Solution. Using the definition of length, dot product and angle in Rn , we have
√ √
|a| = 15, |b| = 21, a · b = −2 − 8 + 0 + 1 = −9,
9 3
cos θ = − p = −√ ,
(15)(21) 35
and hence, on choosing the angle between 0 and π, we have
−3
θ = cos−1 √ = 2.103 . . . radians.
35
♦
Now, in Definition 2 of an angle in Rn , we have assumed that the definition makes sense, i.e., that
the equation
a·b
cos θ =
|a| |b|
can always be solved to obtain a real number as the value for the angle θ. But, since
−1 6 cos θ 6 1 for real numbers, a real solution for θ is possible if and only if
a·b
−1 6 6 1.
|a| |b|
A proof that this inequality is true for all non-zero vectors in Rn is given in the following important
theorem.
Theorem 3 (The Cauchy-Schwarz Inequality). If a, b ∈ Rn , then −|a| |b| 6 a · b 6 |a| |b|.
[X] Proof. Note first that the inequality is clearly true if either a or b is a zero vector.
For b 6= 0, consider
q(λ) = |a − λb|2 for λ ∈ R.
Then, from the properties of lengths and dot products in Rn , we have that q > 0 for all λ ∈ R and
hence that
0 6 q = (a − λb) · (a − λb) = |a|2 − 2λa · b + λ2 |b|2 .
This q(λ) is a quadratic function of λ which has a minimum at
a·b
λ= .
|b|2
The minimum value is
a·b (a · b)2
q0 = q = |a|2 − .
|b|2 |b|2
Now, as q(λ) > 0 for all λ, we have
(a · b)2
0 6 q0 = |a|2 − .
|b|2
c
2019 School of Mathematics and Statistics, UNSW Sydney
50 CHAPTER 2. VECTOR GEOMETRY
Thus,
(a · b)2 6 |a|2 |b|2 ,
and therefore
−|a| |b| 6 a · b 6 |a| |b|,
and the proof is complete.
Another useful inequality which follows immediately from the Cauchy-Schwarz inequality is the
following inequality for lengths of vectors.
|a + b| 6 |a| + |b|.
Proof.
On taking positive square roots of both sides of this Figure 2: The Triangle Inequality.
inequality, we then obtain the result to be proved.
As illustrated in Figure 2, the geometric interpretation of the triangle inequality is that the
sum of two sides of a triangle is greater than or equal to the third side.
Note that the zero vector is orthogonal to every vector, including itself.
On using the formula for the dot product in terms of angles, we see that two vectors are
orthogonal if
a · b = |a| |b| cos θ = 0,
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.3. APPLICATIONS: ORTHOGONALITY AND PROJECTION 51
i.e., if either |a| or |b| or cos θ is zero. As a vector is zero if its length is zero, and cos θ = 0 only
when θ is a right angle, we have the result that vectors a and b are orthogonal if either vector is
the zero vector or if the vectors are at right angles to each other.
Note. Two non-zero vectors at right angles to each other are also said to be perpendicular to
each other or to be normal to each other.
By the definitions of length and orthogonality, the set of standard basis vectors {e1 , . . . , en } is
a set of vectors of unit length at right angles to each other. Sets of vectors with this property are
of great practical importance, and they have been given a special name.
The connection between the dot product and lengths and angles provides a simple test for a set
of vectors to be an orthonormal set.
Example 1. The three standard basis vectors e1 , e2 , and e3 of R3 form an orthonormal set.
Similarly, e1 and e3 are orthogonal since e1 · e3 = 0. Finally, e2 and e3 are orthogonal since
e2 · e3 = 0.
Thus, the three vectors e1 , e2 , e3 are each of unit length and they are mutually orthogonal, and
hence they form an orthonormal set.
Note that a compact form of writing the conditions for an orthonormal set in Rn are that, for
1 6 i 6 n and 1 6 j 6 n, (
0 for i 6= j;
ei · ej =
1 for i = j.
√1 − √13 √1
6
2
Example 2. Show that the three vectors u1 =
0
, u2 = √1 , u3 = √2 form an
3 6
− √12 − √31 √1
6
orthonormal set. Find scalars λ1 , λ2 , λ3 such that e1 = λ1 u1 + λ2 u 2 + λ3 u 3 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
52 CHAPTER 2. VECTOR GEOMETRY
2
1 2 1 2
u1 · u1 = √ + 0 + −√ = 1,
2 2
1 1 1 1 1
u1 · u2 = √ −√ +0× √ + − √ − √ = 0,
2 3 3 2 3
u1 · e1 = u1 · (λ1 u1 + λ2 u2 + λ3 u3 )
1
√ = λ1 (u1 · u1 ) + λ2 (u1 · u2 ) + λ3 (u1 · u3 ),
2
1
we have λ1 = √ . Similarly,
2
1 1
λ 2 = u 2 · e1 = − √ and λ 3 = u 3 · e1 = √ .
3 6
1 1 1
e1 = √ u 1 − √ u 2 + √ u 3 .
2 3 6
The method used in the above example can generally be used to write any vector in the span
of an orthonormal set as a linear combination of vectors in this set.
Example 3. Show that the three altitudes of a triangle are concurrent, i.e., they intersect at a
point.
Solution. As shown in Figure 3, let the three vertices be A, B and C (coordinate vectors a, b,
c), and let D and E be the points at which the altitudes from A and B intersect the opposite sides
of the triangle. Finally, let P (coordinate vector p) be the point of intersection of the altitudes AD
and BE.
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.3. APPLICATIONS: ORTHOGONALITY AND PROJECTION 53
(p − c) · (a − b) = 0,
2.3.2 Projections
−−→ a·b
In both cases, OP = b. Thus we can formally define a projection in terms of the dot
|b|2
product as follows.
c
2019 School of Mathematics and Statistics, UNSW Sydney
54 CHAPTER 2. VECTOR GEOMETRY
The geometric properties in R2 used to motivate this definition can be proved to be true for
projections in Rn also. We have
Proposition 1. projb a is the unique vector λb parallel to the non-zero vector b such that
(a − λb) · b = 0. (#)
a·b
λ= ,
|b|2
Alternative forms of writing the formula for a projection are sometimes useful. These are
b b
projb a = (a · b) b = |a| cos θ b,
b
where b b = b is the unit vector in the direction of b and where θ is the angle between a and b.
|b|
a1
Example 4. Find the projections of a vector a = a2 on the three standard basis vectors e1 ,
a3
e2 , e3 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.3. APPLICATIONS: ORTHOGONALITY AND PROJECTION 55
−4
1
Solution. a · b = 3, |b|2 = 42, and hence projb a = 1 . ♦
14
5
Note that a simple formula for the length of the projection of a on b is
b = |a · b| .
|projb a| = |a · b|
|b|
3 −3
Example 6. Find the length of the projection of 5 on 1 .
1 −4
3 −3 −3
√ 8
Solution. As 5 · 1 = 8, and as 1 = 26, the length of the projection is √ .
1 −4 −4 26
♦
−−→
Note that AB is the line segment joining some point A on the line to the given point B, while
−→
AP is the projection of this line segment on the direction of the line.
An alternative method of solving this problem is to use the fact that
−−→ −−→ −→
P B = AB − AP = b − a − projd (b − a),
−−→
and then the shortest distance is |P B|.
Example 7. Find the distance from the point (2, −1, 3) to the line through the points (0, 1, 4) and
(4, 2, 9).
c
2019 School of Mathematics and Statistics, UNSW Sydney
56 CHAPTER 2. VECTOR GEOMETRY
Solution. Let A, B and C be the points (0, 1, 4), (2, −1, 3) and (4, 2, 9), respectively. Suppose
that P is the foot of the
perpendicular
from B to the line AC.
Referring
to Figure 5, the length of
2 4
−−→ −→
the projection of AB = −2 on the direction d = AC = 1 is
−1 5
−→ |− −→
AB · d| |8 − 2 − 5| 1
AP = =√ =√ .
|d| 16 + 1 + 25 42
−−→
Together with |AB| = 3, the distance from the point to the line is
−−→ r 1
√
377
BP = 32 − = √ .
42 42
♦
a1 x1 + a2 x2 + a3 x3 = 0
b1 x 1 + b2 x 2 + b 3 x 3 = 0
This pair of equations can be easily solved in the usual way to obtain a solution which can be
written in the form
a2 b3 − a3 b2
x = λ a3 b1 − a1 b3 ,
a1 b2 − a2 b1
where λ is a real parameter. This expression for x looks like some kind of “product” of a and b.
a1 b1
Definition 1. The cross product of two vectors a = a2 and b = b2 in
a3 b3
R3 is
a 2 b3 − a 3 b2
a × b = a3 b1 − a1 b3 .
a 1 b2 − a 2 b1
Note that the cross product of two vectors is a vector. For this reason the cross product is
often called the vector product of two vectors, in contrast to the dot product which is a scalar
and is often called the scalar product of two vectors. Note also that the cross product a × b has
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.4. THE CROSS PRODUCT 57
the important property that it is perpendicular to the two vectors a and b. As an exercise you
might like to check directly that a × b is orthogonal to a by checking that a · (a × b) = 0.
There are several tricks available for remembering the formula for a cross product. The most
common trick is to use determinant notation. The more general theory of determinants will be
covered in a later chapter. We define here a 2 × 2 determinant by
a b
c d = ad − bc.
a1 b1
To find the cross product of two vectors a2 and b2 we write these as rows in a 3 × 3
a3 b3
determinant:
e1 e2 e3
a × b = a1 a2 a3 .
b1 b2 b3
To calculate this, we use the following procedure. Firstly, take the vector e1 and multiply it by the
2 × 2 determinant obtained by deleting the row and column in which e1 is contained. That is, we
write
a 2 a 3
e1 .
b2 b3
Then take −e2 and repeat the process, followed by e3 . Each of the 2 × 2 determinants can be found
using the definition above.
We write
e1 e2 e3
a × b = a1 a2 a3
b1 b2 b3
a 2 a 3 a 1 a 3 a 1 a 2
= e1
− e2
+ e3
b2 b3 b1 b3 b1 b2
= e1 (a2 b3 − a3 b2 ) − e2 (a1 b3 − a3 b1 ) + e3 (a1 b2 − a2 b1 )
1 0 0
= 0 (a2 b3 − a3 b2 ) − 1 (a1 b3 − a3 b1 ) + 0 (a1 b2 − a2 b1 )
0 0 1
a2 b3 − a3 b2
= a3 b1 − a1 b3 .
a1 b2 − a2 b1
The determinant is expanded along the first row and as usual e1 , e2 , e3 are the standard basis
vectors of R3 . While this may appear complicated at first, with practice it is much easier than
using the formula for the cross product.
1 4
Example 1. Find the cross product of −2 and −5 .
3 6
c
2019 School of Mathematics and Statistics, UNSW Sydney
58 CHAPTER 2. VECTOR GEOMETRY
2. a × b = −b × a. The cross product is not commutative. If the order of vectors in the cross
product is reversed, then the sign of the product is also reversed.
Proof. Each of the properties listed in Proposition 1 can be proved by expanding each side of the
properties using the definition of cross product given in Definition 1. For example, for Property 2,
we have on expansion that
a2 b3 − a3 b2 b2 a3 − b3 a2
a × b = a3 b1 − a1 b3 = − b3 a1 − b1 a3 = −b × a.
a1 b2 − a2 b1 b1 a2 − b2 a1
There are several useful relations for the cross products of the standard basis vectors e1 , e2 , e3
in R3 .
Proposition 2. The three standard basis vectors in R3 satisfy the relations
1. e1 × e1 = e2 × e2 = e3 × e3 = 0,
2. e1 × e2 = e3 , e2 × e3 = e1 , e3 × e1 = e2 .
Proof. The proof of these relations follows immediately from Property 1 of Proposition 1 and
from Definition 1. For example,
e1 e2 e3 0
e1 × e2 = 1 0 0 = 0 = e3 .
0 1 0 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.4. THE CROSS PRODUCT 59
(e1 × e2 ) × e2 = −e1 ,
but e1 × (e2 × e2 ) = 0.
Proposition 3. Suppose A, B are points in R3 that have coordinate vectors a and b, and ∠AOB =
θ then |a × b| = |a| |b| sin θ.
a2 b3 − a3 b2 2
L.H.S. = a3 b1 − a1 b3
a1 b2 − a2 b1
= a22 b23 + a23 b22 − 2a2 a3 b2 b3 + a23 b21 + a21 b23 − 2a3 a1 b3 b1 + a21 b22 + a22 b21 − 2a1 a2 b1 b2
as required.
c
2019 School of Mathematics and Statistics, UNSW Sydney
60 CHAPTER 2. VECTOR GEOMETRY
−→
a = OA = a1 i
j-direction
−−→ −−→
b = OP + P B = b1 i + b2 j
a × b = a 1 b2 k
B
a×b
O a A P i-direction b
a-direction a
Figure 6: Geometry of a × b.
Hence, we have
a1 b1 0
a = 0 , b = b2 and a × b = 0 ,
0 0 a1 b2
|a × b| = a1 b2 .
Now a1 = |a|, and from trigonometry, b2 = |b| sin θ, and hence |a × b| = |a| |b| sin θ, in agreement
with Proposition 3.
We have therefore shown that:
a × b is a vector of length |a| |b| sin θ in the direction perpendicular to both a and b as given
by the right-hand rule.
This statement is usually taken as the definition of the cross product in physics and engineering
courses.
Note. The above proof is valid since if a, b are arbitrary in R3 then we can apply a rotation to
move a to αi. We then rotate about the x-axis to move b to βi + γj with γ > 0. Since rotations
preserve lengths and angle sizes the results
(a × b) · a = 0, (a × b) · b = 0 and
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.4. THE CROSS PRODUCT 61
2.4.3 Areas
The length of the cross product a × b also has an alternative geometric interpretation in terms of
the area of a parallelogram with sides a and b. Consider the picture of Figure 7.
P B a C
θ
b b
θ
O a A
Figure 7: The Cross Product and the Area of a Parallelogram.
The area of the parallelogram OACB in Figure 7 is “base times perpendicular height”.
−→ −−→
Area = |OA||OP | = |a| |b| sin θ = |a × b|.
The vector a × b is perpendicular to both a and b, so it is normal to the plane of the parallelogram.
Example 2. Find the area of a parallelogram with vertices at points A (1, 0, 1), B (−2, 1, 3), and
C (3, 1, 4).
−3 2
−−→ −→
Solution. One parallelogram can be formed with sides AB = 1 and AC = 1 . The
2 3
1
1 √
−−→ −→
cross product is AB × AC = 13 , and hence the area is 13 = 195. ♦
−5 −5
Note. There are three parallelograms which can be formed from three vertices. The areas of all
three are equal. As an exercise you might like to find the other two parallelograms with the same
three vertices and check that they have the same area.
It is necessary to consider the case of parallel lines and the case of skew (non-parallel) lines sepa-
rately. Be careful, non-parallel lines in R3 may not intersect.
Parallel Lines. The distance between two parallel lines is the same as the distance between a
point on one of the lines and the other line. We only need to use the method in 2.3.3 to find the
distance between a point on one line and the other line.
Skew Lines. The distance, i.e. the shortest distance, between two skew lines is obtained by
drawing a perpendicular to both lines. The direction of the perpendicular is in the direction of the
cross product of the directions of the lines. The shortest distance is the length of the projection on
this perpendicular of a line segment joining any point on one line to any point on the other.
c
2019 School of Mathematics and Statistics, UNSW Sydney
62 CHAPTER 2. VECTOR GEOMETRY
P2
n = d1 × d2 is perpendicular
d1 to d1 and d2
n
−−−→ −−→
P1 P2 = projn AB
d2 P1 B
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.5. SCALAR TRIPLE PRODUCT AND VOLUME 63
In this section we shall briefly examine the properties of the scalar triple product and give it a
geometric interpretation as a volume. Although the vector triple product is useful in physics and
engineering, we shall not consider it any further in this mathematics course.
It is important to notice that in evaluating the scalar triple product, the cross product must
be calculated before the dot product. The expression (a · b) × c has no meaning, because a · b
is a scalar and the cross product of a scalar and a vector has no meaning.
Some properties of the scalar triple product are listed in the following proposition.
Proposition 1. For a, b, c ∈ R3 ,
1. a · (b × c) = (a × b) · c, that is, the dot and cross can be interchanged.
2. a · (b × c) = −a · (c × b), that is, the sign is reversed if the order of two vectors is reversed.
3. a · (a × b) = (a × a) · b = 0, that is, the scalar triple product is zero if any two vectors are
the same.
4. The scalar triple product can be written using the determinant notation.
a 1 a2 a3
a · (b × c) = b1 b2 b3 .
c1 c2 c3
This means that we replace i by a1 , j by a2 and k by a3 in the determinant form of the cross
product.
Proof. The proof of Property 1 follows immediately on using the definitions of the dot and cross
product to expand the two expressions. We have
a1 b2 c 3 − b 3 c 2
a · (b × c) = a2 · b3 c1 − b1 c3
a3 b1 c 2 − c 1 b2
= a1 b2 c3 − a1 b3 c2 + a2 b3 c1 − a2 b1 c3 + a3 b1 c2 − a3 c1 b2 .
a 2 b3 − a 3 b2 c1
(a × b) · c = a3 b1 − a1 b3 · c2
a 1 b2 − a 2 b1 c3
= a2 b3 c1 − a3 b2 c1 + a3 b1 c2 − a1 b3 c2 + a1 b2 c3 − a2 b1 c3 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
64 CHAPTER 2. VECTOR GEOMETRY
The two fully expanded expressions are equal, and hence the result is proved.
Property 2 is an immediate consequence of the fact that b × c = −c × b.
Property 3 follows immediately from Property 1 and from the fact that a × a = 0 for all a.
Property 4 can be proved by expanding both the scalar triple product and the determinant and
noting that the expansions of the two are equal.
Note. Property 3 gives one more proof that the cross product a × b is perpendicular to a. Clearly,
b · (a × b) = 0, and hence b is also orthogonal to a × b.
A
C
a
n=b×c c
O b B
Figure 9: The Scalar Triple Product and the Volume of a Parallelepiped.
The parallelepiped formed from the vectors a, b, and c is called the parallelepiped spanned by
a, b, and c.
From geometry, the volume of a parallelepiped is “area of base times perpendicular height”.
The base is the parallelogram whose sides are the vectors b and c, and hence from the results of
Section 2.4.2, the area of the base is |b × c| and the direction of the perpendicular to the base is
the direction of n = b × c. From the results of Section 2.3.2, the length of the projection of the
vector a on the perpendicular n is the perpendicular height, and hence
|a · n| |a · (b × c)| |a · (b × c)|
Perpendicular height = = = .
|n| |b × c| Area of base
Hence, the volume of the parallelepiped is given by the formula
Volume = |a · (b × c)|.
1 −2 3
Example 1. Find the volume of the parallelepiped spanned by 2 , 4 , and 5 .
3 −1 1
Solution. Using the determinant formula for the scalar triple product, we have
1 −2 3 1 2 3
2 · 4 × 5 = −2 4 −1 = 1(4 + 5) − 2(−2 + 3) + 3(−10 − 12) = −59,
3 −1 1 3 5 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.6. PLANES IN R3 65
2.6 Planes in R3
Another useful application of vectors and dot and cross products is to the geometry of planes in
three dimensions. There are three common forms for the equation of a plane in R3 . These are
“parametric vector form”, “Cartesian form”, and “point-normal form”. Each of these equations
has a direct geometric interpretation. In this section we shall discuss the geometric interpretation
of each of these forms and we shall show how to convert one form to another. We shall also show
how to find the distance between a point and a plane in R3 .
x = c + λ1 v1 + λ2 v2 for λ1 , λ2 ∈ R.
Cartesian Form. In Section 1.5, we have shown that the linear equation in three unknowns,
a1 x1 + a2 x2 + a3 x3 = b,
represents a plane in R3 . This linear equation is also often called the Cartesian form of the
equation of a plane in R3 . We have already shown that this Cartesian form can be converted to a
parametric vector form by solving the linear equation.
It is important to note that a single linear equation a1 x1 +· · ·+an xn = b is the equation of a plane
only if n = 3. For n = 2, the equation is a1 x1 + a2 x2 = b which represents a line. In general, solving
a single linear equation with n unknowns yields a parametric vector form of solution containing
n − 1 parameters, whereas a parametric vector equation of a plane must contain 2 parameters.
The Cartesian form can be given a geometric x3 -axis
interpretation in terms of intercepts on the three
coordinate axes. (0, 0, d3 )
We first divide the Cartesian form by the right
hand side and rearrange it as
x1 x2 x3 (0, d2 , 0)
+ + = 1, O
d1 d2 d3 x2 -axis
b b b
where d1 = , d2 = , d3 = .
a1 a2 a3
Then, to obtain the intercept on the x1 axis, we set (d1 , 0, 0)
x2 = x3 = 0 in the equation, and find x1 = d1 . By x1 -axis
a similar argument, the intercept on the x2 axis is
Figure 10: a1 x1 + a2 x2 + a3 x3 = b.
d2 and the intercept on the x3 axis is d3 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
66 CHAPTER 2. VECTOR GEOMETRY
The rule for obtaining intercepts is therefore to rewrite the equation with 1 on the right, and then
the intercepts are the reciprocals of the coefficients of the variables. Notice that if the coefficient
of x1 is zero, then the plane is parallel to the x1 axis. A similar result applies if the coefficient of
x2 or x3 is zero.
Example 1. Write down the equation of a plane which has the intercepts 4, 7, −2 on the three
coordinate axes.
Solution. The coefficients of the variables are the reciprocals of the intercepts, and hence the
equation is
x1 x2 x3
+ + = 1, or 14x1 + 8x2 − 28x3 = 56.
4 7 −2
♦
n1 x1 + n2 x2 + n3 x3 = n1 c1 + n2 c2 + n3 c3 = b,
which is the Cartesian form of the equation of a plane. Thus, the equation n · (x − c) = 0 is the
equation of a plane. As for the Cartesian form, the point-normal equation is the equation of a
plane for three-dimensional vectors only.
The name “point-normal form” for the equation n · (x − c) = 0 is based on the following
geometric interpretation (see Figure 11). Clearly, x = c is a solution of the equation, and hence c
is the position vector of a point on the plane. If x is any other point in the plane, then the line
segment x − c lies in the plane. The equation then says that the vector n is normal to all line
segments lying in the plane. n is called a normal to the plane. The point-normal form is therefore
the most convenient form of equation to use when a point on the plane and a normal to the plane
are known.
P
n x −c
x
c
O
Figure 11: Point-Normal Form n · (x − c) = 0.
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.6. PLANES IN R3 67
Example 4 (Conversion from Cartesian to point-normal form). Find the point-normal form of the
equation
3x1 − 7x2 + 5x3 = 21.
Solution. A comparison of the Cartesian and point-normal forms shows that the coefficients
of x1
, x2 ,
x3 are just the coordinates of a normal to the plane. Thus, a normal to the plane is
3
n = −7 .To find some point on the plane, we let x2 = x3 = 0, and then from the equation we
5
21
find x1 = = 7. Hence (7, 0, 0) is a point on the plane.
3
3 7
A point-normal form is therefore −7 · x − 0 = 0. ♦
5 0
Example 5 (Conversion from parametric vector to point-normal form). Find the point-normal
form for the parametric vector equation
3 0 −2
x = −5 + λ1 3 + λ2 6 .
1 4 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
68 CHAPTER 2. VECTOR GEOMETRY
0 −2
Solution. The plane is parallel to 3 and 6 . The cross product of these two vectors is
4 1
0 −2 −21
therefore normal to the plane. Thus, a normal to the plane is n = 3 × 6 = −8 . As
4 1 6
−21 3
(3, −5, 1) is a point on the plane, a point-normal form is −8 · x − −5 = 0. ♦
6 1
x2 does not appear in the equation, but don’t forget it. As in 1.5.3, we set x2 = λ1 to be a real
parameter. Also, x3 can have any value in the equation, and hence we set x3 = λ2 to be a second
real parameter. Then, we obtain x1 = −2 + 2x3 = −2 + 2λ2 . Then, on rewriting the solution in
vector form, we obtain a parametric vector form of the equation of the plane to be
x1 −2 + 2λ2 −2 0 2
x = x2 = λ1 = 0 + λ1 1 + λ2 0 .
x3 λ2 0 0 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.6. PLANES IN R3 69
Example 8. Find a point-normal form of the equation of a plane through the three points
A (3, 1, 2), B (0, −2, 1), and C (1, 2, 3).
Solution. The plane is parallel to the two line segments
−3 −2
−−→ − →
AB = −3 and AC = 1 .
−1 1
Hence a normal to the plane is
−3 −2 −2
−−→ −→
n = AB × AC = −3 × 1 = 5 .
−1 1 −9
A (3, 1, 2) is a point on the plane, and hence a point-normal form is
−2 3
5 · x − 1 = 0.
−9 2
♦
Note. Although we have not stated it explicitly, most of the solutions given to the examples in
this section are not unique. For example, any multiple of a normal to a plane is still a normal, any
multiple of a vector parallel to a plane is still a vector parallel to the plane, there are an infinite
number of points on a plane etc. However, the intercepts of a plane on the coordinate axes are
unique.
c
2019 School of Mathematics and Statistics, UNSW Sydney
70 CHAPTER 2. VECTOR GEOMETRY
Example 9. Find the distance between the point (2, −1, 3) and the plane
0 4 −3
x = 4 + λ1 1 + λ2 1 .
2 2 5
4 −3 3
Solution. (0, 4, 2) is a point on the plane and n = 1 × 1 = −26 is a vector
2 5 7
normal
to the
plane. A vector
from the point (0, 4, 2) on the plane to the given point (2, −1, 3) is
2 0 2
−1 − 4 = −5 .
3 2 1
2
The shortest distance is the length of the projection of −5 on n, which is
1
2
−5 · n
1 143
=√ .
|n| 734
♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
2.7. GEOMETRY AND MAPLE 71
giving its Cartesian equation or four points on it or the end-points of a diameter or its center and
its radius. To display the specifications of one of these things you need to use detail (see the
example below). For a plane, the detail includes a Cartesian equation for the plane. (If you want
to find a normal to a plane p use
NormalVector(p);)
Be warned that if you specify a plane or sphere by means of an equation then Maple will want
you to specify the names of the variables which are associated with the three axes. You can do this
by listing them as a third argument to the plane or sphere command, as in
plane(P,x+y+z=1,[x,y,z]);
If you leave out the [x,y,z] then Maple will, rather strangely, prompt you to enter the name of
the x-axis, to which you reply x;, and similarly for the other two axes.
When you have set up objects of these types you can, for example, use the command distance
to find the distance between two of them or the command intersection to find the intersection
of two of them (except the intersection of a line and a sphere) or the command FindAngle to find
the angle between two of them. You can use the Maple help to find out more about any of these
commands and to find out about the many other commands available in geom3d.
In the following example, we first label the points A(0, 1, 2) and B(2, 3, 1) and
the line AB
2
through A and B. Applying detail to AB shows that the direction of the line is 2 and the
−1
line can be expressed in parametric vector form as
0 2
x = 1 + t 2 , t ∈ R.
2 −1
Then we assign the label P to the plane through C(4, 5, 6) with normal (1, 1, 1) and use detail to
find that P can be described by the Cartesian equation
x + y + z = 15.
Then we assign the label X to the point of intersection of the line AB and the plane P and find
that the coordinates of X are (8, 9, −2). Finally, we find that the plane ABC through the three
points A, B, C can be described by the Cartesian equation
12x − 12y + 12 = 0.
Notice that we use a colon : to suppress the output of most of the commands because the output
would just be an echo of assigned names.
with(geom3d):
point(A,[0,1,2]),point(B,[2,3,1]):
line(AB,[A,B]):
detail(AB);
Warning, assume that the name of the parameter in the parametric equations is t
Warning, assuming that the names of the axes are x, y, and z
name of the object: AB
form of the object: line3d
equation of the line: [ x = 2* t, y = 1+2* t, z = 2- t]
c
2019 School of Mathematics and Statistics, UNSW Sydney
72 CHAPTER 2. VECTOR GEOMETRY
point(C,[4,5,6]):
plane(P,[C,[1,1,1]]):
detail(P);
Warning, assuming that the names of the axes are x, y and z
name of the object: P
form of the object: plane3d
equation of the plane: -15+ x+ y+ z = 0
intersection(X,AB,P):
detail(X);
name of the object: X
form of the object: point3d
coordinates of the point: [8, 9, -2]
plane(ABC,[A,B,C]):
detail(ABC);
Warning, assuming that the names of the axes are x, y and z
name of the object: ABC
form of the object: plane3d
equation of the plane: 12+12* x-12* y = 0
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 2. VECTOR GEOMETRY 73
2. [H] Find the cosines of the internal angles of the triangles whose vertices have the following
coordinate vectors:
4 6 5 0 −1 3
a) A 0 , B 2 and C 1 ;
b) A 2 , B 3 and C 1 ;
2 1 6 1 0 2
1 0 −2
−2 4 1
c) A
0 , B −2 and C 0 .
3 5 3
3. [R] A cube has vertices at the 8 points O (0, 0, 0), A (1, 0, 0), B (1, 1, 0), C (0, 1, 0), D (0, 0, 1),
E (1, 0, 1), F (1, 1, 1), G (0, 1, 1). Sketch the cube, and then find the angle between the
−−→ −→
diagonals OF and AG.
6. [H] Use the dot product to prove that the diagonals of a square intersect at right angles.
c
2019 School of Mathematics and Statistics, UNSW Sydney
74 PROBLEMS FOR CHAPTER 2
8. [H] Consider the triangle ABC in R3 formed by the points A(3, 2, 1), B(4, 4, 2) and C(6, 1, 0).
4 2
−2 1 −1
c) the projection of 2 on the direction of the line x = 0 + λ 1 .
7 2 2
11. [X] A point P in Rn has coordinate vector p. Find the coordinate vector of the point Q which
is the reflection of P in the line ℓ which passes through the point a parallel to the direction
d.
NOTE. Define Q to be the point which lies in the same plane as P and ℓ with ℓ bisecting
the interval P Q.
a·b
a) Show q(λ) is a minimum when λ = λ0 = .
|b|2
b) Determine q(λ0 ) and hence show that −|a| |b| 6 a · b 6 |a| |b|.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 2. VECTOR GEOMETRY 75
13. [X] Let B be a point in Rn with coordinate vector b. Let x = a + λd, λ ∈ R be the equation
of a line. Do the following:
a) Show that the square of the distance from B to an arbitrary point x on the line is
given by
q(λ) = |b − a|2 − 2λ(b − a) · d + λ2 |d|2 .
b) Find the shortest distance between the point B and the line by minimising q(λ).
c) If P is the point on the line closest to B, show that
−−→
P B = b − a − projd (b − a),
−−→
and show that P B is orthogonal to the direction d of the line.
NOTE. This problem proves that the shortest distance between a point and a line is
obtained by “dropping a perpendicular from the point to the line”.
14. [R] Find the cross product a × b of the following pairs of vectors:
0 1 3 −2
a) a = 2 and b = 3 ,
b) a = 1 and b = 6 ,
−4 2 4 1
1 2
c) a = 9 and b = 0 .
2 −5
1 −2
15. [R][V] Find a vector which is perpendicular to 3 and 0 .
2 4
16. [H] Prove the following properties of cross products for vectors a, b, c ∈ R3 :
a) a × a = 0; b) a × b = −b × a;
c) a × (λb) = λ(a × b); d) a × (b + c) = a × b + a × c.
17. [R] Find the areas of, and the normals to the planes of, the following parallelograms:
1 0
a) the parallelogram spanned by 3 and 2 ;
2 4
b) a parallelogram which has vertices at the three points A (0, 2, 1), B (−1, 3, 0) and
−→ −→
C (3, 1, 2) and sides AB and AC.
c
2019 School of Mathematics and Statistics, UNSW Sydney
76 PROBLEMS FOR CHAPTER 2
18. [R][V] Find the areas of the triangles with the following vertices:
21. [X] Let a, b, c be three vectors in R3 which satisfy the relations b = c × a and c = a × b.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 2. VECTOR GEOMETRY 77
a) Find a parametric vector equations of the line through A and B and the line through
C and D.
b) Find the shortest distance between the lines AB and CD.
c) Find the point P on AB and point Q on CD such that P Q is the shortest distance
between the lines AB and CD.
27. [R][V] Find parametric vector, point-normal, and Cartesian forms for the following planes:
−1
a) the plane through (1, 2, −2) perpendicular to 1 ;
2
−1 2
b) the plane through (1, 2, −2) parallel to 1 and 3 ;
2 1
c) the plane through the three points (1, 2, −2), (−1, 1, 2) and (2, 3, 1);
d) the plane with intercepts −1, 2 and −4 on the x1 , x2 and x3 axes;
c
2019 School of Mathematics and Statistics, UNSW Sydney
78 PROBLEMS FOR CHAPTER 2
−1
e) [X] the plane through (1, 2, −2) which is parallel to 2 and the line of intersection
−2
of the planes
1 2 1
x · 2 = 0 and x = λ1 1 + λ2 0 .
3 2 −1
Let Π be the plane through A and parallel to the lines OB and OC.
31. [R] Let P be the plane in R3 through the points A = (1, 2, 0), B = (0, 1, 2), and C = (−1, 3, 1).
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 2. VECTOR GEOMETRY 79
32. [X] a) Let a and v be two non-zero vectors in R3 . Show how to write v as c + d where c is
parallel to a and d is perpendicular to a.
b) Consider the plane
x x 2 1 0
Π = y : y = 1 + λ1 2 + λ2 1 , λ1 , λ2 ∈ R
z z 0 1 −1
1
and the vector v = 1 . By using a) (or otherwise) express v as c + d where d is
1
parallel to Π and c is perpendicular to Π. (We call d the projection of v onto Π).
c
2019 School of Mathematics and Statistics, UNSW Sydney
80 CHAPTER 2. VECTOR GEOMETRY
81
Chapter 3
COMPLEX NUMBERS
The main purpose of this chapter is to introduce the system of complex numbers. In the calculus
part of this subject we concentrate on the set R of real numbers which we think of as corresponding
to the points on the number line.
−2 −1 0 1 x 2
b
x is real. x ∈ R.
If one wants to solve all quadratic equations then the real numbers do not suffice. In particular,
if − 4ac < 0 then it is not possible to find real solutions to ax2 + bx + c = 0. In this chapter,
b2
we will construct a larger set of numbers in which such an equation can be solved. Indeed, in this
larger set, every polynomial equation has at least one solution. Much of mathematics becomes
simpler (not more complicated) when complex numbers are used and applications to areas such as
physics, chemistry, electrical and mechanical engineering, oceanography, economics and the theory
of dynamical systems are made simpler by using complex numbers.
c
2019 School of Mathematics and Statistics, UNSW Sydney
82 CHAPTER 3. COMPLEX NUMBERS
together with rules of addition and multiplication. This set of numbers has the property that
addition or multiplication of natural numbers always produces another natural number, whereas
subtraction or division may not. Thus 3 − 5 and 35 are not natural numbers. We say that the set
of natural numbers is closed under the operations of addition and multiplication, whereas the set
is not closed under the operations of subtraction or division.
A very limited class of equations have solutions in N. For example, in N, x+3 = 7 and 5x+2 = 17
can be solved, but x + 7 = 3 and 5x + 2 = 18 can not! Thus, to solve all linear equations, a larger
set of numbers is required.
A set of numbers that is closed under subtraction (i.e., for which subtraction is always possible)
can be obtained by extending the natural number system by introducing a new number (−1).
Then, after using the usual rules for addition and multiplication, the set of integers
Z = {. . . , −2, −1, 0, 1, 2, . . .}
is obtained.
However, the set Z is not closed under division as, for example, 53 is not an integer. At this stage
x + 7 = 3 has a unique solution but 5x + 2 = 18 still has no solution. To solve such an equation,
we need fractions. Thus, we extend the system of integers to the set of rational numbers, Q,
defined by
p
Q= : p, q ∈ Z for q 6= 0 .
q
This system is then closed under the four standard operations of arithmetic of addition, subtraction,
multiplication and division (division by zero excluded).
Now that the rationals are in the set that we are focusing on, all equations in one variable with
rational coefficients, such as 5x + 2 = 18 or 23 x + 41 = 1 can be solved. All solutions will be rational
although sometimes a solution may happen to be an integer. Indeed, if we now consider the general
equation in one variable ax + b = c with a, b, c ∈ Q then this has a unique solution x = (c − b)/a
unless a = 0.
The rationals are the first and primary example of a mathematical concept called a field. (See
definition 1.) A field is a set (of numbers) which satisfy “twelve number laws”. These laws, or
axioms as they are called, form a minimal list of properties that one needs in order to be able to
add, subtract, multiply and divide (by non-zero numbers).
Much elementary mathematics can be carried out using rational numbers, as is done by your
calculator.
√ √ The set of real numbers, R, contains all the rationals,
√ √ √along with numbers such as
2, 3, π, e, etc. which are not rational. The proofs that 2, 3, 5, . . . are irrational are very
straightforward. The proofs that π and e are irrational are harder. You will see a proof of the
irrationality of e later in the year. The last question in the HSC Extension 2 2003 paper outlines
a proof of the irrationality of π.
Using real numbers, we can now solve
√ some quadratic and higher degree equations. For example, √
x2 − 3 = 0 has 2 real solutions x = ± 3 and x3 − x2 − 3x + 3 = 0 has 3 real solutions x = 1, ± 3.
The set of real numbers also satisfies all twelve number laws and thus forms a field. Hence all
equations ax + b = c with a, b, c ∈ R, a 6= 0, have a unique solution.
Note that the general motivation behind the development of number systems sketched above
is that some set of numbers is extended to a new set by introducing new “numbers” so that some
operation (e.g., subtraction, division, or finding lengths of sides of squares) is always possible in
the extended set.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.2. INTRODUCTION TO COMPLEX NUMBERS 83
Definition 1. Let F be a non-empty set of elements for which a rule of addition (+)
and a rule of multiplication are defined. Then the system is a field if the following
twelve axioms (or fundamental number laws) are satisfied.
c
2019 School of Mathematics and Statistics, UNSW Sydney
84 CHAPTER 3. COMPLEX NUMBERS
A complex number written in the form a + bi, where a, b ∈ R, is said to be in Cartesian form.
The real number a is called the real part of a + bi, and b is called the imaginary part. The set C
contains all the real numbers (when b = 0). Numbers of the form bi, with b real (b 6= 0), are called
purely imaginary numbers. The set of complex numbers also satisfies the twelve number laws,
and so it also forms a field.
Example 1. Some examples of complex numbers in Cartesian form are
π π
3 + 4i, 2 − i = 2 + (−1)i, −5i = 0 + (−5)i, 6 = 6 + 0i and cos + i sin .
3 3
♦
z + w = (a + c) + (b + d)i.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.3. THE RULES OF ARITHMETIC FOR COMPLEX NUMBERS 85
Since i2 = −1, this can be simplified to (ac − bd) + (bc + ad)i. Hence we define the product, zw, by
(It is wise to multiply out the terms producing real numbers first and then purely imaginary
numbers second.)
Division. To divide two complex numbers we use a similar process to that used in “rationalising
the denominator”. For example,
√ √
1 (2 − 3) 2− 3 √
√ = √ √ = = 2 − 3,
2+ 3 (2 + 3)(2 − 3) 4−3
c
2019 School of Mathematics and Statistics, UNSW Sydney
86 CHAPTER 3. COMPLEX NUMBERS
Note. The formula for division by w = c + id given above fails if and only if c2 + d2 = 0. But,
since c and d are real, c2 + d2 = 0 if and only if c = 0 and d = 0, that is, if and only if w = 0. Thus
the formula for complex division fails if and only if the denominator is 0.
Before concluding this section on complex number arithmetic, we should point out that we now
have three examples of fields; namely, the rational numbers Q, the real numbers R and the complex
numbers C.
Proposition 1. [X] The following properties hold for addition of complex numbers:
Proposition 2. [X] The following properties hold for multiplication of complex numbers:
One final point should be made. So far we have stressed the similarities between real number
arithmetic and complex number arithmetic. However, there are also many important differences.
One important difference is that while it makes sense to say that a real number is positive or that
one real number is greater than (or less than) another, it does not make sense to say that a complex
number is positive or that one complex number is greater than (or less than) another. That is,
complex numbers cannot be ordered.
NOTE.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.4. REAL PARTS, IMAGINARY PARTS AND COMPLEX CONJUGATES 87
So far, we handle complex numbers by writing them in a + bi form. With the above properties,
we can write the real and imaginary parts of a complex number in terms of the complex number
and its conjugate. We can also prove some general results in a simpler way.
c
2019 School of Mathematics and Statistics, UNSW Sydney
88 CHAPTER 3. COMPLEX NUMBERS
z+w
Example 5. Let z, w ∈ C such that zz = ww. Prove that is purely imaginary.
z−w
z+w z+w
Proof. To show that is purely imaginary, we only need to show that Re is 0. Using
z−w z−w
α + α = 2Re(α),
!
z+w 1 z+w z+w
Re = +
z−w 2 z−w z−w
1 z+w z+w
= +
2 z−w z−w
(z + w)(z − w) + (z − w)(z + w)
=
2(z − w)(z − w)
zz − zw + wz − ww + zz + zw − wz − ww
=
2(z − w)(z − w)
2(zz − ww)
=
2(z − w)(z − w)
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.5. THE ARGAND DIAGRAM 89
Example 1. Plot the numbers 4, −4, i, −2i, −2 + 3i, 3 − 4i, −3 − 4i on an Argand diagram. The
solution is shown in Figure 2. ♦
Imaginary axis
−2 + 3i
b
b
i
−4 4
b b
0 Real axis
b
−2i
−3 − 4i 3 − 4i
b b
Figure 2: Plots of 4, −4, i, −2i, −2 + 3i, 3 − 4i, −3 − 4i, on the Argand Diagram.
Note that we can write either i and −2i on the y-axis above, or simply mark 1 and −2,
remembering that this is the imaginary axis.
Plotting real and imaginary parts leads to a simple geometric picture of addition and subtraction
of complex numbers. Addition and subtraction is done by adding or subtracting x-coordinates and
y-coordinates separately, as shown in Figure 3.
Imaginary axis
Imaginary axis
z1 + z2b
z2 b y2 b
z2
z1 b
z1
b
x2
c
2019 School of Mathematics and Statistics, UNSW Sydney
90 CHAPTER 3. COMPLEX NUMBERS
Imaginary axis
An alternative representation for complex
numbers, which proves to be very useful, is
obtained by using plane polar coordinates r
and θ instead of the Cartesian coordinates x y z = x + yi
and y. The coordinate r is the distance of
a point from the origin, and θ is an angle r
measured from the positive x-axis, as shown
in Figure 4. θ
Take a complex number z 6= 0, then from 0 x Real axis
Figure 4, Pythagoras’ theorem gives
p
r = x2 + y 2 with r > 0. Figure 4: Polar Coordinates of a Complex
Number.
By trigonometry,
x x y y
cos θ = p = and sin θ = p = .
x2 + y 2 r x2 + y 2 r
Thus the relations between the real and imaginary parts of z = x + yi and the polar coordinates r
and θ are
Re(z) = x = r cos θ and Im(z) = y = r sin θ,
and hence a complex number z 6= 0 can be written using the polar coordinates r and θ as:
It is important to note here that the angle θ for a given complex number z = x + yi is not uniquely
defined; since adding or subtracting 2π produces exactly the same values for x and y and hence
the same complex number z. This result is summarised in the following proposition.
x1 = r1 cos θ1 = r2 cos θ2 = x2
y1 = r1 sin θ1 = r2 sin θ2 = y2
and z1 = z2 .
Conversely, if z1 = z2 , then we have x1 = x2 and y1 = y2 . Hence,
q q
r1 = x1 + y1 = x22 + y22 = r2 ,
2 2
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.6. POLAR FORM, MODULUS AND ARGUMENT 91
and since r1 , r2 6= 0,
x1 x2 y1 y2
cos θ1 = = = cos θ2 and sin θ1 = = = sin θ2 ,
r1 r2 r1 r2
The polar coordinate r that we have associated with a complex number is often called the
modulus of the complex number. The formal definition is:
The quantity |z| is also called the magnitude of z or the absolute value of z. Note that it has
a geometric interpretation as the distance r = |z| of the point z from the origin in an Argand
diagram. Note also that zz = |z|2 .
p
Example 1. | − 4| = 4, |2i| = 2, and |3 − 4i| = 32 + (−4)2 = 5. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
92 CHAPTER 3. COMPLEX NUMBERS
θ
θ
0 Real axis 0 Real axis
0 0
Real axis θ Real axis
θ
b b
z z
We can see from the diagrams, that if z lies in the first or second quadrant, we measure the
principal argument θ anticlockwise. For z in the 3rd or the 4th quadrant, we measure θ, as a
negative angle, clockwise. Note that we leave Arg(0) undefined.
One useful strategy is always to draw a diagram, then use the tangent ratio to find the acute
angle α formed by the corresponding triangle and use this to find the principal argument.
√ √ √
Example 2. Find the arguments of 1 + i, −1 + 3 i, − 3 − i and 1 − 3 i.
Solution.
Imaginary
We first plot z = 1 + i on an Argand diagram axis
as in Figure 6. The complex number lies in the 1st b
1+i
quadrant, and so 0 < Arg(z) < π/2. As shown in
the figure, 1
θ
tan θ = 1, and
0 1 Real axis
π
Arg (1 + i) = θ = .
4
Figure 6.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.6. POLAR FORM, MODULUS AND ARGUMENT 93
√ Imaginary
From Figure 7, the complex number −1 + 3i √
lies in the 2nd quadrant. Also, −1 + 3i b
axis
√ π √
tan α = 3, so α = . 3 θ
3
α
Hence
1 0 Real axis
√ 2π
Arg −1 + 3 i = θ = π − α = .
3
Figure 7.
√ Imaginary
From Figure 8, the complex number − 3−i lies
axis
in the 3rd quadrant. Also,
√
1 π 3 0
tan α = √ , so α = . α Real axis
3 6 1
b
θ
Hence √
√ − 3−i
5π
Arg 3 − i = −π + α) = − .
6
Figure 8.
√ Imaginary
From Figure 9, the complex number 1− 3 i lies axis
in the 4th quadrant. Also
√ π 1
tan α = 3, so α = . 0 Real axis
3 α
√
Hence, 3
√ π
Arg 1 − 3 i = −α = − .
b
√
3 1−i 3
Figure 9.
♦
In the special case that a given complex number is real or purely imaginary, we can easily read
the principal argument from an Argand diagram.
Example 3. Find the principal argument for each of the numbers 4, −4, i and −2i.
Solution. From Figure 2 in Section 3.5, we can easily read the arguments as
π π
Arg(4) = 0, Arg(−4) = π, Arg(i) = , Arg(−2i) = − .
2 2
♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
94 CHAPTER 3. COMPLEX NUMBERS
π
Example 4. Find the “a + ib” form of the complex number with modulus 4 and argument − .
6
Solution.
√ !
π π 3 1 √
z = 4 cos − + i sin − =4 − i = 2 3 − 2i.
6 6 2 2
♦
and
sin(θ1 + θ2 ) = cos θ1 sin θ2 + sin θ1 cos θ2 ,
the result follows.
Lemma 1 can be used to derive a very important and useful theorem for integer powers of
complex numbers in polar forms. This theorem is called De Moivre’s Theorem.
Theorem 2 (De Moivre’s Theorem). For any real number θ and integer n
Proof. We shall prove this theorem by proving that the condition (#) holds for the four separate
cases of n > 0, n = 0, n = −1 and n < −1, and hence that it holds for all n ∈ Z.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.7. PROPERTIES AND APPLICATIONS OF THE POLAR FORM 95
CASE 2. n = 0. The condition (#) is true for this case, provided we use the convention that
z 0 = 1 for any complex number z.
1
CASE 3. n = −1. By definition, z −1 = . Then, applying the division rule for complex numbers
z
to z = cos θ + i sin θ, we have
1
(cos θ + i sin θ)−1 =
cos θ + i sin θ
1 cos θ − i sin θ
= ×
cos θ + i sin θ cos θ − i sin θ
cos θ − i sin θ
=
cos2 θ + sin2 θ
= cos(−θ) + i sin (−θ),
De Moivre’s Theorem provides a simple formula for integer powers of complex numbers. How-
ever, it can also be used to suggest a meaning for complex powers of complex numbers. To make
this extension to complex powers it is actually sufficient just to give a meaning to the exponential
function for imaginary exponents. We first make the following definition.
This definition may appear somewhat arbitrary at first, but there are several reasons why it is
c
2019 School of Mathematics and Statistics, UNSW Sydney
96 CHAPTER 3. COMPLEX NUMBERS
reasonable. First recall that for real constants a, θ and φ, we have for integer n,
(eaθ )n = eanθ , (1)
e0 = 1, (3)
d aθ
e = aeaθ . (4)
dθ
In fact, properties (3) and (4) are often taken as a definition of the exponential function for real
numbers.
The next point to notice is that if a is replaced by i and if eiθ is replaced by (cos θ + i sin θ)
in these four formulae, then all four formulae are still satisfied. In fact, equation (1) would be
De Moivre’s Theorem, equation (2) would be lemma 1, equation (3) would obviously be true, and
equation (4) would be
d
(cos θ + i sin θ) = i(cos θ + i sin θ) = − sin θ + i cos θ,
dθ
which is also true (provided we assume that differentiation of expressions containing the symbol i
can be carried out in the same way as if i were a real constant). Thus, cos θ + i sin θ has exactly
the same properties that we would like eiθ to have, and so the definition
eiθ = cos θ + i sin θ
is consistent with our experience with other exponential functions.
Note also that, since cosine is an even function and sine is an odd function, so
e−iθ = cos(−θ) + i sin(−θ) = cos θ − i sin θ,
which is the conjugate of eiθ .
We should also say that in more sophisticated treatments of complex functions, eiθ will be
defined by other means (usually a “power series”) and Euler’s formula then becomes a theorem.
De Moivre’s Theorem and Euler’s formula have a wide variety of uses, ranging from calculation of
powers of complex numbers to calculation of roots of complex numbers to derivation of trigonometric
formulae etc. We shall examine some of these applications in the remainder of this chapter.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.7. PROPERTIES AND APPLICATIONS OF THE POLAR FORM 97
(Note: At High School you may have taken r(cos θ + i sin θ) as the polar form and abbreviated it
to r cis (θ). This notation is not used in this course.)
Since eiθ = ei(θ+2kπ) for all k an integer, it is sometimes convenient to express the polar form using
a general argument, that is,
Using Euler’s formula, we can rewrite the equality proposition for polar forms (Proposition 1 of
Section 3.6) as
z1 = r1 eiθ1 = r2 eiθ2 = z2
if and only if r1 = r2 and θ1 = θ2 + 2kπ for k ∈ Z.
The polar form is very useful for multiplication and division of complex numbers. The formulae
for multiplication and division of polar forms are:
where k is an integer, chosen so that −π < Arg(z1 z2 ) 6 π. That is, the rule is to multiply the
moduli and add the arguments. For division, we have
z1 r1 |z1 | z1
= = and Arg = Arg(z1 ) − Arg(z2 ) + 2kπ,
z2 r2 |z2 | z2
where k is an integer chosen so that −π < Arg(z1 /z2 ) 6 π. That is, the rule is to divide the moduli
and subtract the arguments.
√
Example 2. Use √ polar forms to find the modulus and argument of (−1 − i)(1 − i 3) and
(−1 − i)/(1 − i 3).
√
Solution. The modulus and argument of −1 − i and 1 − i 3 are
√ 3π
| − 1 − i| = 2 and Arg (−1 − i) = − ,
4
√ √ π
|1 − i 3| = 2 and Arg 1 − i 3 = − .
3
c
2019 School of Mathematics and Statistics, UNSW Sydney
98 CHAPTER 3. COMPLEX NUMBERS
where k = 0 has been chosen to obtain a principal argument in the interval (−π, π]. ♦
The next example shows a reasonably simple method for finding the square roots of a complex
number in Cartesian form. It rests on the observation that if z ∈ C and z = a + bi with a, b ∈ R,
then
|z 2 | = |z · z| = |z| |z| = |z|2 or |(a + bi)2 | = a2 + b2 ,
and also Re(z 2 ) = a2 − b2 , Im(z 2 ) = 2ab.
Solution. We want to find all complex solutions to z 2 = −5 − 12i. Writing z = a + bi, this is
equivalent to finding all real solutions (a, b) to the equation (a + bi)2 = −5 − 12i. Expanding the
left-hand side and equating real and imaginary parts, we have
a2 − b2 = −5
2ab = −12.
Also, p
a2 + b2 = |(a + ib)2 | = | − 5 − 12i| = (−5)2 + (−12)2 = 13.
So we want to solve the system of equations
a2 + b2 = 13
a2 − b2 = −5
2ab = −12.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.7. PROPERTIES AND APPLICATIONS OF THE POLAR FORM 99
Solving the first pair in this system, we get a2 = 4 and b2 = 9. This means that a = ±2 and b = ±3.
The third equation is satisfied precisely when we choose opposite signs for a and b. Therefore the
two square roots of −5 − 12i are
z = 2 − 3i and − 2 + 3i,
or more compactly,
z = ±(2 − 3i).
♦
z 2 − (4 + i)z + (5 + 5i) = 0.
Solution. We solve this using the quadratic formula. The proof of the quadratic formula uses
only field axioms and the fact of the existence of square roots and thus carries over directly to
quadratic polynomials with complex coefficients. The details are left as an exercise.
Proceeding, the solutions to z 2 − (4 + i)z + (5 + 5i) = 0 are
p
4 + i ± (4 + i)2 − 4(5 + 5i)
z=
√ 2
4 + i ± −5 − 12i
=
2
From the previous example, the square roots of −5 − 12i are ±(2 − 3i). Thereby the roots to
the quadratic are
4 + i + (2 − 3i) 4 + i − (2 − 3i)
z= =3−i and z= = 1 + 2i
2 2
We can easily check that z = 3 − i and z = 1 + 2i are roots to the equation by substitution. ♦
Example 5. Show that the set of numbers of unit modulus, that is, the set
S = {z ∈ C : |z| = 1},
as |z1 | = 1 and |z2 | = 1. Thus z1 /z2 ∈ S, and hence S is closed under division. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
100 CHAPTER 3. COMPLEX NUMBERS
Example 6. [X] Suppose a and b are real numbers (not both zero) and w = (az + b)/(bz + a).
Show that, if |z| = 1, then |w| = 1.
Solution. We have
az + b |az + b|
|w| = = .
bz + a |bz + a|
Now, |z| = 1, so z = cos θ + i sin θ, θ real.
p
|az + b| = |a(cos θ + i sin θ) + b| = (a cos θ + b)2 + (a sin θ)2
p
= a2 + 2ab cos θ + b2
p
and |bz + a| = |b(cos θ + i sin θ) + a| = a2 + 2ab cos θ + b2 .
So |w| = 1. ♦
z = reiθ ,
z n = rn einθ .
√ 10
Example 7. Calculate 1 + i 3 .
√ π
Solution. For z = 1 + i 3, we have |z| = 2, Arg(z) = 3 and so z = 2eiπ/3 . Hence,
√ 10 iπ/3 10 10 10iπ/3 10 −2iπ/3 10 2π 2π √
1+i 3 = 2e =2 e =2 e =2 cos − i sin = 29 −1 − i 3 .
3 3
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.7. PROPERTIES AND APPLICATIONS OF THE POLAR FORM 101
If z0 6= 0 the nth roots of z0 can be found by equating the polar forms of z n and z0 . Thus, if
z = reiθ and z0 = r0 eiθ0 , we have
z n = rn einθ = r0 eiθ0 ,
Thus
1/n θ0 + 2kπ θ0 2kπ
r = r0 and θ= = + for k ∈ Z,
n n n
and
1 θ0 +2kπ
iθ i
z = re = r0 e n n
. (*)
As k ranges over the integers, z takes precisely n different values. These n values can be found
by letting k take any n consecutive values. Thus z0 has precisely n distinct nth roots.
Imaginary axis
Hence, the fifth roots of unity are complex
numbers, z, given by b
z 5 = e2kπi
b
for k ∈ Z,
2π
5 b
By (*) above, these roots are 2π
5 Real axis
i( 0+2kπ ) b
z=e 5 ,
b
Polar forms for the five fifth roots of unity are therefore
♦
Note that these solutions occur in conjugate pairs, eg. e2πi/5 and e−2πi/5 and so on.
As shown in Figure 10, these fifth roots are equally spaced out on a circle of radius 1 with angles
2π
between them.
5
Example 9. Find all sixth roots of −2.
c
2019 School of Mathematics and Statistics, UNSW Sydney
102 CHAPTER 3. COMPLEX NUMBERS
Imaginary axis
Hence, the sixth roots of −2 are complex
numbers z, such that b
z 6 = 2e(π+2kπ)i for k ∈ Z. b
π b
3
So, the sixth roots are
π Real axis
1
i
(2k+1)π b 3 b
√ ! √ ! √ ! √ !
1 3+i 1 1 3−i 1 3+i 1 1 3−i
−2 6 , −2 i, 2
6 6 , 2 6 , 2 i, −2
6 6
2 2 2 2
There is again a simple geometric picture of this result. From figure 11, the 6 roots lie on a
1 2π π
circle of radius 2 6 with angles = between them.
6 3
In summary, there are always exactly n nth roots of a non-zero complex number. On the
1 2π
Argand diagram, these nth roots of z0 lie on a circle of radius |z0 | n with angles between them.
n
n n!
where the numbers = are the binomial coefficients.
k k!(n − k)!
For small values of n the binomial coefficients may be easily calculated using Pascal’s triangle.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.8. TRIGONOMETRIC APPLICATIONS OF COMPLEX NUMBERS 103
n BINOMIAL COEFFICIENTS
0 1
1 1 1
2 1 2 1
3 1 3 3 1
4 1 4 6 4 1
5 1 5 10 10 5 1
Note that each coefficient (except the 1’s at the end) is obtained by adding the two coefficients
immediately above it.
Example 1. For n = 4, the coefficients are 1, 4, 6, 4, 1, and hence
Then, on using i2 = −1, i3 = −i, i4 = 1, and separately equating real and imaginary parts on the
left and right hand sides gives
and
sin 4θ = 4 cos3 θ sin θ − cos θ sin3 θ .
Note that the formula for cos 4θ can be rewritten in term of powers of cos θ only by using
2
sin θ = 1 − cos2 θ to replace the sine terms by cosines. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
104 CHAPTER 3. COMPLEX NUMBERS
On first adding and then subtracting these formulae, we obtain the important formulae
1 inθ 1 inθ
cos nθ = e + e−inθ , sin nθ = e − e−inθ .
2 2i
In particular, we have
1 iθ 1 iθ
cos θ = e + e−iθ , sin θ = e − e−iθ .
2 2i
We can apply the above formulae to derive trigonometric formulae which relate powers of sin θ
or cos θ to sines or cosines of multiples of θ.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.9. GEOMETRIC APPLICATIONS OF COMPLEX NUMBERS 105
Solution. We use the fact that cos θ is the real part of eiθ . The required sum is then the real part
of the sum
Sn = eiθ + e2iθ + · · · + eniθ .
k
Since ekiθ = eiθ , the sum Sn is a geometric progression with ratio of successive terms given by
R = eiθ . Then, on using the formula for the sum of a geometric progression (see Section 3.11), we
have
1 − Rn 1 − einθ
Sn = R + R2 + · · · + Rn = R 1 + R + · · · + Rn−1 = R = eiθ .
1−R 1 − eiθ
We require the real part of Sn . The simplest way of finding the real part is to use the following
trick. Note that
θ
iθ iθ/2 −iθ/2 iθ/2 iθ/2
1−e =e e −e = −2ie sin ,
2
and hence that
inθ/2 sin nθ nθ
iθ e 2 i(n+1)θ/2 sin 2
Sn = e θ
=e θ
.
eiθ/2 sin 2 sin 2
♦
Sums of this type are used in X-ray diffraction, solid-state physics, chemistry, signal processing
in electrical engineering, and tomography, as well as in many other areas where periodic functions
and waves must be analysed.
c
2019 School of Mathematics and Statistics, UNSW Sydney
106 CHAPTER 3. COMPLEX NUMBERS
Imaginary axis
z
b
w| z
− b
y−b
|z
w α α
b b
x−a w
Imaginary axis
w b
α
wb
z b b
The fact that |z −w| is the distance between the points z and w and that Arg (z − w) is the angle
between the arrow from w to z and a line in the direction of the positive real axis are important
and should be remembered. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.9. GEOMETRIC APPLICATIONS OF COMPLEX NUMBERS 107
Imaginary axis
4+ib
4+ib
An alternative way of arriving at a geometric interpretation of a set is to use the “x + yi” (or
Cartesian) form of a complex number. For example, using z = x + yi, we can square the equation
|z − 4 − i| = 3 and rewrite it to obtain
|z − 4 − i|2 = |(x − 4) + (y − 1)i|2 = (x − 4)2 + (y − 1)2 = 9,
which corresponds to the equation of a circle with centre at x = 4 and y = 1 and of radius 3. ♦
n π πo
Example 3. Sketch the set 6 Arg (z − 4 − i) 6
z∈C:− .
3 4
Solution. The above set is the set of all complex numbers z such that
π π
− 6 Arg(z − 4 − i) 6 .
3 4
Imaginary axis
c
2019 School of Mathematics and Statistics, UNSW Sydney
108 CHAPTER 3. COMPLEX NUMBERS
Solution. The given set is the set of all complex numbers z which satisfy both |z − 3| > 2 and
Re(z) 6 4. The inequality |z − 3| > 2 describes the region on and outside the circle of radius 2
centred at 3, while Re(z) = x 6 4 corresponds to the half-plane on and to the left of the line x = 4.
The required set is the intersection of these two regions as both inequalities must be satisfied. The
plots are shown in Figure 17. In practise, only the last diagram need to be shown.
b b b b
0 3 0 4 0 3 4
Solution. The inequality |z − 1 + i| < 2 represents the open disc of radius 2 with centre at 1 − i.
0 0 0
b b
1−i 1−i
To indicate that the circle |z − 1 + i| = 2 is not included in the set, we draw it with dashes.
Also, Im(z) = y > 0 describes the half-plane on and above the real axis. The required set is the
union of these two regions (as either inequality being satisfied puts the point in the set). ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.10. COMPLEX POLYNOMIALS 109
Note. If a0 , a1 , . . . , an are real and z takes only real values, then we say that the polynomial is
defined over R.
Theorem 1 (Remainder Theorem). The remainder r which results when p(z) is divided by z − α
is given by r = p(α).
The major difference between polynomials over the complex numbers and polynomials over the
real numbers is contained in the following theorem.
Theorem 3 (The Fundamental Theorem of Algebra). A polynomial of degree n > 1 has at least
one root in the complex numbers.
c
2019 School of Mathematics and Statistics, UNSW Sydney
110 CHAPTER 3. COMPLEX NUMBERS
We shall not try to prove the fundamental theorem here. The proof is usually given in courses
on functions of a complex variable.
There are several important points to note about the fundamental theorem.
The theorem is not true in general for polynomials over R. For example, the real quadratic q
defined by q(z) = 1 + z 2 does not have any real roots. In contrast, it has the two complex roots
±i.
We can combine the Fundamental Theorem of Algebra and the Factor Theorem to prove the
following extremely important theorem.
Theorem 4 (Factorisation Theorem). Every polynomial of degree n > 1 has a factorisation into
n linear factors of the form
where the n complex numbers α1 , α2 , . . . , αn are roots of p and where a is the coefficient of z n .
Thus,
p(z) = (z − α1 )p1 (z) = a(z − α1 )(z − α2 ) . . . (z − αn+1 ).
Hence, (#) is true for polynomials of degree n + 1 whenever it is true for polynomials of degree
n, and as (#) is true for n = 1, it is true for all n > 1, by induction.
Finally, note that in the factorisation shown in (#) a is the coefficient of the z n term in p.
p1 (z) = (z + 1)2 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.10. COMPLEX POLYNOMIALS 111
The polynomial
√ p3 (z) does not easily factor. Using the quadratic formula, the roots are
−1 ± i 3
z = , so
2
1 √ 1 √
p3 (z) = z − −1 + i 3 z− −1 − i 3 .
2 2
♦
The Factorisation Theorem guarantees that a polynomial of degree n always has n roots, but it
does not tell us how to actually find these roots. For polynomials of degree n 6 4, exact formulae
for the roots in terms of the coefficients have been found. For quadratics, the exact formula for
the roots is well known and very easy to use. For cubic polynomials (degree 3) the exact formula
is called Cardano’s formula and it is also reasonably easy to use. For quartic polynomials (degree
4) there is also an exact formula for the roots but the formula is very complicated. However, for
polynomials of degree n > 4, there are no formulae for the roots in terms of square roots, cube
roots, etc. A proof of this fact is given in courses on Galois Theory.
In general, it is either difficult or impossible to find exact roots, and hence an exact factorisation,
for higher degree polynomials. It is usually necessary to resort to approximate numerical methods
to find the roots of a polynomial. At the present time the best general-purpose numerical method
is based on finding the “eigenvalues” of a “companion matrix” for the polynomial. These numerical
methods are discussed in advanced courses on numerical matrix algebra.
There are, however, some very simple types of higher degree polynomials for which exact roots,
and hence an exact factorisation, can be found. One such case is for polynomials of the form
p(z) = z n − a, since in this case the solutions of
zn − a = 0 or equivalently of z n = a,
are the nth roots of the number a. This problem has been discussed in Section 3.7.3.
Example 3. Factorise z 6 + 1.
Solution. We first solve z 6 +1 = 0 or z 6 = −1 by finding the sixth roots of −1. Since the modulus
and argument of −1 are | − 1| = 1 and Arg(−1) = π, we have
z 6 = e(π+2kπ)i for k ∈ Z.
So, the sixth roots in polar forms are
(2k+1)π
z = ei 6 , for k = −3, −2, −1, 0, 1, 2.
−i 5π −i π2 −i π6 i π6 i π2 5π
That is, e 6,e ,e , e , e , and ei 6 .
By the Factor Theorem, we have
z 6 + 1 = z − eiπ/2 z − e−iπ/2 z − eiπ/6 z − e−iπ/6 z − e5iπ/6 z − e−5iπ/6 .
An alternative form can be obtained by putting the roots into the “a + ib” form. You should do
this as an exercise, and you should obtain the result
6 1 √ 1 √ 1 √ 1 √
z +1 = (z−i)(z+i) z − 3+i z− 3−i z− − 3+i z− − 3−i .
2 2 2 2
♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
112 CHAPTER 3. COMPLEX NUMBERS
An immediate consequence of this proposition is that the roots of a complex polynomial with real
coefficients are either real or occur in conjugate pairs. This fact can be used to obtain a factorisation
of a polynomial with real coefficients into linear or quadratic factors with real coefficients.
Proposition 6. If p is a polynomial with real coefficients, then p can be factored into linear and
quadratic factors all of which have real coefficients.
Proof. For a polynomial with real coefficients, if α is a root which is not real then so is α, and
hence both z − α and z − α are factors of p. On multiplying these factors, we obtain
(z − α)(z − α) = z 2 − (α + α)z + αα.
Then, from section 3.3, we have that
α + α = 2Re(α), and αα = |α|2 ,
where Re(α) and |α|2 are both real numbers. Hence,
q(z) = (z − α)(z − α) = z 2 − 2Re(α) z + |α|2
is a quadratic with real coefficients. Thus, all complex factors can be replaced in pairs by quadratic
factors with real coefficients, and the proof is complete.
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.11. APPENDIX: A NOTE ON PROOF BY INDUCTION 113
Example 4. Factorise z 5 + 32 into linear and quadratic factors with real coefficients.
Solution. We first solve z 5 = −32 to find the fifth roots of −32. Using the usual procedure, we
obtain the five solutions
Note that this factorisation is certainly not obvious. If you multiply out the right hand side and
π 3π
compare coefficients you will find some very surprising relations between cos and cos . ♦
5 5
1 − rn+1
1 + r + r2 + · · · + rn = ,
1−r
where r 6= 1 is a fixed number. In this example, P (n) gives the formula for the sum of n + 1 terms
of a geometric progression with ratio r. ♦
Example 3. For each integer n > 2, let P (n) be the proposition that “n can be completely factored
into prime numbers”. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
114 CHAPTER 3. COMPLEX NUMBERS
As we shall see, all the propositions P (n) in examples 1, 2 and 3 are true.
There are several versions of “proof by induction”, but the two most commonly used are based
on what are sometimes called the first and second principles of induction.
First Principle of Induction. Let n0 ∈ Z, and let P (n), for n > n0 , be propositions. Then, if
1. P (n0 ) is true, and
Example 2. We prove the formula for the sum of a geometric progression, i.e., that:
For n ∈ N and r 6= 1,
1 − rn+1
1 + r + · · · + rn = .
1−r
Proof. We first note that the proposition P (0) states that
1−r
1= ,
1−r
which is clearly true for r 6= 1.
Now, if P (n) is true for some fixed integer n > 0, then the sum of n + 2 terms is
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.12. APPENDIX: THE BINOMIAL THEOREM 115
2. for each n > n0 , P (n + 1) is true whenever all propositions P (m) are true
for m = n0 , . . . , n (i.e. n0 6 m 6 n),
Proof. The proposition P (n) is that n can be factored into primes. P (2) then asserts that 2 can
be factored into primes. This result is clearly true, since 2 is itself a prime.
We now show that, if P (m) is true for all integers m with 2 6 m 6 n, then P (n + 1) is also
true.
Now, the integer n + 1 must either be prime or not prime.
Case 1. n + 1 is prime. Then n + 1 is already factored and hence P (n + 1) is true in this case.
Case 2. n + 1 is not prime. Then, there are two integers m1 and m2 less than n + 1 such that
n + 1 = m1 m2 . But, as P (m) is true for 2 6 m 6 n, m1 and m2 can both be factored into primes,
and hence n + 1 can also be factored into primes. Thus, P (n + 1) is true in this case also.
We have shown that P (2) is true, and that P (n+1) is true whenever P (m) is true for all m with
2 6 m 6 n. Hence, from the second principle of induction, P (n) is true for all integers n > 2.
In fact the two apparently different types of induction are the same, as we can easily show.
Simply let the proposition Q(n) be “P (m) is true for m = n0 , . . . , n”. Then use the first type of
proof.
n(n − 1) n−2 2
(a + b)n = an + nan−1 b + a b + · · · + nabn−1 + bn
2!
Xn
n n−k k
= a b ,
k
k=0
Proof. The proof of the theorem is based on the first principle of induction. In this case, the
proposition P (n) to be proved true is that the formula given above for (a + b)n is correct.
For n = 1, P (1) asserts that (a + b)1 = a1 + b1 , which is clearly true.
c
2019 School of Mathematics and Statistics, UNSW Sydney
116 CHAPTER 3. COMPLEX NUMBERS
Now, if the formula is correct for some integer n > 1, then we have
On multiplying out, we find that the coefficient of an+1 and bn+1 are 1 and that the coefficient
of an+1−k bk is
n n
+ .
k k−1
Now,
n n n! n!
+ = +
k k−1 k!(n − k)! (k − 1)!(n − k + 1)!
n!
= (n − k + 1 + k)
k!(n − k + 1)!
(n + 1)! n+1
= = .
k!(n − k + 1)! k
Hence,
Xn
n+1 n+1 n + 1 n+1−k k
(a + b) =a + a b + bn+1
k
k=1
X n + 1
n+1
= an+1−k bk .
k
k=0
c
2019 School of Mathematics and Statistics, UNSW Sydney
3.13. COMPLEX NUMBERS AND MAPLE 117
c
2019 School of Mathematics and Statistics, UNSW Sydney
118 PROBLEMS FOR CHAPTER 3
Questions marked with [R] are routine, [H] harder, [M] Maple and [X] are for MATH1141
only. You should make sure that you can do the easier questions before you tackle the more
difficult questions. Questions marks with a V have video solutions available on Moodle.
2. [R] Is the set {0, 1, 2, 3, 4, 5, 6, 7, 8, 9} closed under addition? Prove your answer.
3. [H] Can any finite set of integers be closed under addition? Prove your answer.
8. [R][V] Use the quadratic formula to find all complex roots of the following polynomials.
a) z 2 + z + 1, b) z 2 + 2z + 3, c) z 2 − 6z + 10,
d) − 2z 2 + 6z − 3, e) z 4 + 5z 2 + 4.
√ √ 3
9. [H] Show that 3+1 + 3 − 1 i = 16(1 + i).
√ √ 2 √
10. [R] Simplify 3 + 4i + 3 − 4i (where we assume z has non negative real part).
2 2
a + bi a − bi
11. [H][V] Simplify − where a and b are real numbers not both zero.
a − bi a + bi
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 3. COMPLEX NUMBERS 119
2−i 1
12. [R] Find Re(z), Im(z) and z for z = −1 + i, 2 + 3i, 2 − 3i, , .
1 + i (1 + i)2
z
13. [R] Let z = 1 + 2i and w = 3 − 4i. Calculate z 2 and , expressing the answers in Cartesian
w
form.
16. [R] Prove that for any two complex numbers z and w
1
a) Im(z) = (z − z) b) 2Re(z) = z + z
2i
1 1
c) (z − w) = z − w d) = ,
z z
z z
e) zw = z w f) = .
w w
17. [H] a) Use the properties of the complex conjugate to show that if the complex number α
is a root of a quadratic equation ax2 + bx + c = 0 with a, b, c being real coefficients,
then so is α.
b) Write down the monic quadratic polynomial with real coefficients which has 3 − 2i as
one of its roots.
c) Does the result of a) generalise to higher degree polynomials?
18. [R][V] Find the modulus, principal argument and polar form of each of the following numbers
and plot them on an Argand diagram:
√ −1 i
a) 6 + 6i, b) − 4, c) 3 − i, d) √ − √ , e) − 7 + 3i.
2 2
19. [R] If z = 4 + 3i and w = 2 + i find |3z − 3iw|, Im (1 − i)z − 3|w| .
20. [H] If z = 1 + i, calculate the powers z j for j = 1, 2, . . . , 10 and plot them on an Argand
diagram. Is there a pattern? What is the smallest positive integer n such that z n is a real
number?
21. [R] Find the “a + ib” form of the complex numbers whose moduli and principal arguments are
c
2019 School of Mathematics and Statistics, UNSW Sydney
120 PROBLEMS FOR CHAPTER 3
π 5π
a) |z| = 3, Arg(z) = ; b) |z| = 3, Arg(z) = ;
3 6
2π π
c) |z| = 3, Arg(z) = − ; d) |z| = 3, Arg(z) = − ;
3 6
π
e) [H] |z| = 3, Arg(z) = .
8
22. [R][V] a) Show that z z = |z|2 . Hence, or otherwise, show that if |z| = 1, then z = z −1 .
b) Show that |z| = |z| for all z ∈ C.
c) If z = r(cos θ + i sin θ), show that a polar form for the complex conjugate is
z = r (cos(−θ) + i sin(−θ)).
1−z
23. [H] Show that Re = 0 for any complex z with |z| = 1.
1+z
24. [H] Use zz = |z|2 to prove the identity |z1 + z2 |2 + |z1 − z2 |2 = 2(|z1 |2 + |z2 |2 ).
and deduce that |1 − zw|2 = |z − w|2 if either z or w lies on the unit circle.
w
27. [R] Let z = (1 − i) and w = 2eiπ/3 . Calculate w6 , z − w and and express your answers in
z
Cartesian form.
28. [R] For z = 3e−5πi/6 and w = 1 + i, find Re iw + z 2 .
33. [R] Find the polar, and hence also the Cartesian form for:
1002 √ !−8
√ 5 −1 + i 1 + 3i
a) 3+i , b) √ , c) .
2 2
35. [H] a) Explain why multiplying a complex number z by eiθ rotates the point represented by
z anticlockwise about the origin, through an angle θ.
b) The point represented by the complex number 1 + i is rotated anticlockwise about
π
the origin through an angle of . Find its image in polar and Cartesian form.
6
c) Find the complex number (in Cartesian form) obtained by rotating 6−7i anticlockwise
3π
about the origin through an angle .
4
π
36. [H] If z = reiθ , 0 6 θ 6 , show that
2
√ π
a) (1 − i)z 2 = 2r2 b) Arg (1 − i)z 2 = 2θ − ,
4
√ !
1 + i √3 2 1+i 3 π
c) = , d) Arg = − θ.
z r z 3
38. [R] Find the seventh roots of −1 and plot the roots on an Argand diagram.
39. [R] Find the sixth roots of i and plot the roots on an Argand diagram.
√
40. [H] Find the fifth roots of 16 − 16i 3 and plot the roots on an Argand diagram.
42. [H][V] Show that if ω is an nth root of unity (ω 6= 1 and n > 1) then
ω + ω 2 + · · · + ω n = 0.
43. [X] Show that the set {z ∈ C : |z| 6 1} is closed under multiplication. Is the set closed under
division (zero excluded)? Is the set closed under addition or subtraction?
c
2019 School of Mathematics and Statistics, UNSW Sydney
122 PROBLEMS FOR CHAPTER 3
44. [X] Use the properties of complex conjugates to show that if a, b ∈ R and |z| = 1, then
|a + bz| = |az + b|. Hint: You might find the results of Question 22 useful.
az + bz −1
45. [X] Suppose a and b are real numbers (not both zero) and w = . Show that if
bz + az −1
|z| = 1, then |w| = 1.
Hint: You might find the results of Question 22 useful.
a) Show that π
−1
2 + 2 tan s
for s61
Arg(z) =
3π
− + 2 tan−1 s for s > 1.
2
b) Describe geometrically what happens to z as s increases from −∞ to ∞.
π
48. [H] Suppose θ, φ 6= (2k + 1) where k is an integer. Use the fact that
2
1+z
z=
1 + z −1
a) to find the real and imaginary parts of
1 + cos 2θ + i sin 2θ
;
1 + cos 2θ − i sin 2θ
49. [X] For n > 1, let ω1 , ω2 , ..., ωn be the n distinct nth roots of 1 and let Ak be the point on the
Argand diagram which represents ωk . Let P represent any point z on the unit circle, and
let P Ak denote the distance from P to Ak .
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 3. COMPLEX NUMBERS 123
n
X
b) Deduce that (P Ak )2 = 2n.
k=1
c) Now let P represent the point x on the real axis, −1 < x < 1, prove that
n
Y
P Ak = 1 − xn .
k=1
50. [R][V] Using De Moivre’s theorem and the binomial theorem, prove the identity
cos 3θ = 4 cos3 θ − 3 cos θ.
51. [R] a) Use De Moivre’s Theorem to express cos 6θ and sin 6θ in terms cos θ and sin θ.
b) Write cos 6θ in terms of cos θ only
52. [H] Express cos 7θ and sin 7θ in terms of powers of cos θ and sin θ.
53. [R] a) Derive a formula for cos θ in terms of eiθ and e−iθ .
b) Deduce a formula for cos6 θ in terms of cos kθ, 1 6 k 6 6.
Z π
2 5π
c) Show that cos6 θ dθ = .
0 32
54. [R][V] Express sin5 θ and cos4 θ in terms of sines or cosines of multiples of θ, and hence find
their integrals.
55. [X] a) Use De Moivre’s Theorem to express cos 5θ as a polynomial p(x) in x = cos θ.
b) Put θ = 36◦ = π5 and show that x = cos π5 is a root of P (x) = 16x5 − 20x3 + 5x + 1.
c) Check that P (x) = (x + 1)(4x2 − 2x − 1)2 .
d) What are the 5 roots of P (x)? Give full reasons for your answer.
e) Deduce that cos π5 + cos 3π 7π 9π π 3π 7π 9π 1
5 + cos 5 + cos 5 = 1 and cos 5 cos 5 cos 5 cos 5 = 16 .
56. [X] Let ω1 , ω2 , . . . , ωn be the n distinct nth roots of unity (n > 1). Show that if k is an integer
then
ω1k + ω2k + · · · + ωnk
equals 0 or n. Find the values of k for which the sum is n.
Hint: Write the roots in polar form and sum the resulting geometric progression. See
Example 5 of Section 3.8.
57. [X] Show that if θ is not a multiple of 2π, then the imaginary part of
1 − ei(n+1)θ sin 12 (n + 1)θ sin( 12 nθ)
is .
1 − eiθ sin 12 θ
Hint. See Example 5 of Section 3.8.
c
2019 School of Mathematics and Statistics, UNSW Sydney
124 PROBLEMS FOR CHAPTER 3
60. [R][V] Sketch the set of points on the complex plane corresponding to each of the following:
π 2π
a) |z − i| 6 2, b) |z − i| 6 2 or − 6 Arg (z − i) 6 ,
3 3
c) |z| > 2 and |Im(z)| 6 3, d) Re(z) > Im(z),
π π
e) |z − i| = |z + i|, f) |z − 1 − i| < 1 and − < Arg (z − 1 − i) 6 ,
4 2
g) |z − i| = 2|z + i|, h) [X] |z − i| + |z + i| = 6.
61. [R] Sketch the following on two carefully labelled Argand diagrams.
63. [R] Let z be a complex number. Prove that |z − Re(z)| 6 |z − x| for all real numbers x. Draw
a sketch to illustrate the result.
a) Sketch the subset of the complex plane defined by w = eiα for −π < α 6 π.
b) Given that Arg(z) = θ, prove that |z − eiθ | 6 |z − eiα | for all α ∈ R.
c) Give a geometric interpretation of the result in part b).
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 3. COMPLEX NUMBERS 125
65. [R] Use the remainder theorem to find the following remainders when.
a) 2 + 3z − z 2 + 6z 3 is divided by z − 5,
b) 1 − 6z + 5z 2 − 8z 3 + 2z 4 is divided by z + 2,
c) 3z + 2z 2 + z 3 is divided by z − 1 − i.
66. [R][V] Use the remainder theorem and the factor theorem to show that z − 2 is a factor of
p(z) = 30 − 17z − 3z 2 + 2z 3 . Then divide p by z − 2 and hence find all linear factors of p.
67. [R] Use the method of the previous question to show that z − 1 and z + 2 are factors of
p(z) = −8 − 6z + 7z 2 + 6z 3 + z 4 . Then find all linear factors of p.
69. [H] a) Factorise x8 − 1 into real linear and real quadratic factors.
b) Repeat for x6 + 8.
c
2019 School of Mathematics and Statistics, UNSW Sydney
126 PROBLEMS FOR CHAPTER 3
75. [H] Consider f (t) = t6 + t5 − t4 − 5t3 − 6t2 − 6t − 4. Given that −1 + i is a root of f and that
f also has two real integer roots,
a) factorise f into complex linear factors,
b) factorise f into linear and quadratic factors with real coefficients.
76. [H][V] Let f (z) = z 5 − 2z 4 + 2z 3 − 5z 2 + 10z − 10. Given that 1 + i is a root, find all solutions
to f (z) = 0.
1 + z + z2 + z3 + z4 + z5 + z6 + z7 + z8
78. [X] Let p(z) = 3z − z 3 + 5z 4 + z 6 . You are told that the six roots of p, say α1 , . . . , α6 , are
distinct.
a) Prove that at least two of these roots are real.
b) Show that
α1 + α2 + α3 + α4 + α5 + α6 = 0.
c) Hence or otherwise, show that there is at least one root with positive real part, and
at least one root with negative real part.
d) Show that if |z| > 3, then
|3z − z 3 + 5z 4 | < |z 6 |.
e) Hence or otherwise show that for j = 1, . . . , 6, |αj | 6 3.
79. [X] Cardan (approximately 1545) gave a formula for the roots of the cubic equation
x3 + ax + b = 0
u= − + + v= + +
2 2 27 2 2 27
and where the cube roots u and v must be selected to satisfy uv = a3 . It can be shown
that there are three pairs of values of u and v which satisfy the above conditions.
If a 6= 0 a simpler way of writing Cardan’s formula is that the three roots of the cubic are
of the form
a
x=u− ,
3u
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 3. COMPLEX NUMBERS 127
where u satisfies s
b b 2 a3
u3 = − + + .
2 2 27
a) Use the simpler version of Cardan’s formula to find all three roots of x3 − 6x + 4 = 0.
(Note that complex numbers are used in the calculation even though all three roots
are real).
b) Use the fact that one root of the cubic x3 −6x+4 is 2 to factor the cubic as (x−2)q(x)
where q(x) is a quadratic. Hence find all roots of the cubic. Hence deduce that
√ √
5π −1 + 3 11π 1+ 3
cos = √ and cos =− √ .
12 2 2 12 2 2
4b 4b
cos 3θ = provided − 1 6 6 1.
k3 k3
d) Use the method outlined above to find the three real roots of x3 − 6x − 4 = 0.
where the all coefficients a0 , a1 , . . . , an are integers. Show that, if r/s is a rational root of
p for which the integers r and s have no common factors, then r is a divisor of a0 and s is
a divisor of an . Hence find all rational roots of
82. [X] Let M, N be positive integers. If xM (1 − x)N is divided by (1 + x2 ), and the remainder is
√ (2M − N )π √ (2M − N )π
ax + b, show that a = ( 2)N sin and b = ( 2)N cos .
4 4
83. [R][V] Prove by mathematical induction that for all positive integers n,
1
1.2 + 2.3 + · · · + n(n + 1) = n(n + 1)(n + 2).
3
c
2019 School of Mathematics and Statistics, UNSW Sydney
128 PROBLEMS FOR CHAPTER 3
Use the second Principle of Induction to prove that an 6 1 for all n > 1.
89. [X] Suppose we draw n lines in the plane with no three lines concurrent and no two lines
parallel. Let sn denote the number of regions into which these lines divide the plane. For
example, s1 = 2, s2 = 4, s3 = 7, . . . . Prove that sn+1 = sn + (n + 1). Deduce by induction
that sn = 12 n(n + 1) + 1.
91. [M] Use Maple to evaluate (5 + i)4 (239 − i). Then use de Moivre’s Theorem to show that
π
= 4 cot−1 5 − cot−1 239.
4
c
2019 School of Mathematics and Statistics, UNSW Sydney
129
Chapter 4
One glass lemonade (Why ca’n’t you drink water, like me?)
three sandwiches (They never put in half mustard enough.
I told the young woman so, to her face . . . )
and seven biscuits. Total one-and-twopence.
Lewis Carroll, A Tangled Tale.
Linear equations and matrices are very important because they are used as mathematical models
in virtually all areas in which mathematics is applied in the modern world and because they
also appear at the heart of computational algorithms for solving a vast array of quite diverse
mathematical problems.
The processes described in this chapter are appropriate for the general theory of handling
systems of linear equations. They work well for most small-scale examples.
An understanding of linear equations and matrices is essential for much of the later work in
algebra in this book so you must master the material in this chapter.
c
2019 School of Mathematics and Statistics, UNSW Sydney
130 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Since these equations correspond to straight lines in the plane, case (1) represents two distinct
parallel lines; case (2), two non-parallel lines and in case (3), the two lines are the same.
Now consider some examples of two equations in three unknowns. Assume that none of the
equations is of the form 0x1 + 0x2 + 0x3 = b. The solutions of such simultaneous equations can be
interpreted as the points of intersection (if any) of two planes in R3 . From geometry, we expect
that two planes either
3. are parallel with all points in common, (that is, they are the same plane).
We convert each system to another equivalent system, i.e. a system which has the same solution
set, until we get one which gives us a simple form of the solution set.
x1 + x2 + x3 = 5 (1)
3x1 + 4x2 + 7x3 = 20 (2)
Solution. To solve this pair, we first eliminate x1 from equation (2) by subtracting 3 times
equation (1) from equation (2). The new equation (2) is then
x2 + 4x3 = 5 (2′ )
x1 + x2 + x3 = 5 (1)
x2 + 4x3 = 5 (2′ )
Equation (2′ ) has infinitely many solutions which can be represented using a parameter. Put
x3 = λ, then
x2 = 5 − 4λ.
Substituting this back to equation (1), we have
x1 + (5 − 4λ) + λ = 5,
so x1 = 3λ.
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.1. INTRODUCTION TO LINEAR EQUATIONS 131
The geometric interpretation of this example is that equations (1) and (2) each represents a
plane, and the solutions of system represent the points of intersection of the two planes. In this
particular example, the intersection of thetwo planes is a line. This line passes through the point
3
(0, 5, 0) and is parallel to the vector −4 . ♦
1
Solution. As usual, we eliminate x1 from equation (2). On subtracting 2 × equation (1) from
equation (2), we obtain the new equation (2) as
which clearly has no solution. Thus the system has no solution. The geometric interpretation is
that the planes represented by equations (1) and (2) are parallel and do not intersect. ♦
x1 − 3x2 + x3 = 20 (1)
2x1 − 6x2 + 2x3 = 40. (2)
Solution. Eliminating x1 from equation (2) gives the new equation (2) as
x1 − 3x2 + x3 = 20 (1)
0x1 + 0x2 + 0x3 = 0 (2′ )
Thus the solutions to (1) and (2) are just the solutions to (1) and conversely. There are infinitely
many solutions, but here two of the unknowns need to be specified in order to obtain the third.
Thus two parameters λ1 , λ2 need to be introduced. Put x2 = λ1 and x3 = λ2 , then
x1 = 20 + 3λ1 − λ2 ,
c
2019 School of Mathematics and Statistics, UNSW Sydney
132 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Hence the system has an infinite number of solutions. The geometric interpretation of this example
is that equations (1) and (2) represent the same plane and the ‘solution’ is just a parametric vector
form of expression for that plane. ♦
Now, what about a system of more than two equations in two or three variables? Let us
exclude the cases that one or more of the equations in the system is of the form 0x1 + 0x2 = b or
0x1 + 0x2 + 0x3 = b. In such cases, the system is either has no solutions or is equivalent to a system
of two or less equations.
For a system of three equations in two variables, we can interpret the three equations as three
lines in a plane. Geometrically, we have three cases.
1. The three lines are concurrent, so the system has a unique solution.
2. The three lines do not have a point in common, so the system has no solution. (See Figure 1
for cases in which the three lines are distinct.)
3. The three equations represent the same line. The system has infinitely many solutions.
For a system of three equations in three variables, the three equations represent three planes in
three dimensional space. Geometrically, we have four cases.
1. The three planes intersect at one point, so the system has a unique solution.
2. The three planes do not have a point in common, so the system has no solution. (See Figure 2
for cases in which the three planes are distinct.)
3. The three planes intersect in a line, so there are infinitely many solutions.
4. The three equations represent the same plane. Again, the system has infinitely many solu-
tions.
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.2. SYSTEMS OF LINEAR EQUATIONS AND MATRIX NOTATION 133
Note carefully the position of the subscripts in (∗ ): aij is the coefficient of the variable xj in
the ith equation, that is, the equation index is first and the variable index is second.
A solution to a system of equations is the set of values of the variables which simultaneously
satisfy all the equations. We normally write a solution in form of a column vector in Rn. For
α1
..
instance, if x1 = α1 , . . . , xn = αn satisfy the equations simultaneously, the vector . is a
αn
solution. A system of equations is said to be consistent if it has at least one solution. Otherwise,
the system is said to be inconsistent.
In particular, the system is called homogeneous when all the bi ’s are zero.
Example 1.
2x1 + 3x2 − x3 + x4 = 6
x1 − 4x2 + 5x3 = 2
x1 − 6x3 = 5
x1 + 2x2 − x4 = 7
is a system of 4 linear equations in 4 unknowns. ♦
The variables x1 , x2 , . . . , xn in the system (∗ ) are really just place marker for the coefficients.
Hence it is convenient to adopt a shorthand notation for (∗ ) which records only the coefficients.
This is called the augmented matrix for the system (∗ ):
a11 a12 · · · a1n b1
a21 a22 · · · a2n b2
.. .. .. .. .. .
. . . . .
am1 am2 · · · amn bm
c
2019 School of Mathematics and Statistics, UNSW Sydney
134 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
We denote the matrix of coefficients by A, the column of the right hand side as the vector b
and the column vector of the variables as x. That is
a11 a12 · · · a1n b1 x1
a21 a22 · · · a2n b2 x2
A= . .. .. .. , b = . . and x= .
.. . . . .
. ..
am1 am2 · · · amn bm xn
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.2. SYSTEMS OF LINEAR EQUATIONS AND MATRIX NOTATION 135
This is our first meeting with “matrix multiplication”, which we shall investigate in more detail in
Chapter 5.
Finally we can also write the system of linear equations in vector equation or vector form.
a11 a12 a1n b1
a21 a22 a2n b2
x1 . + x2 . + · · · + xn . = . .
.
. .
. . ..
.
am1 am2 amn bm
x1 a1 + x2 a2 + · · · + xn an = b,
where
a1j b1
a2j b2
aj = . for 16j6n and b = . .
.
. ..
amj bm
Notice that the vector aj contains the coefficients of the variable xj in the system of linear equations.
The vector b is called the right-hand-side vector.
This vector form of the system of equations shows that solving the system corresponds to
expressing the right-hand-side vector b as a linear combination of the vectors a1 , a2 , . . . , an .
Example 3. The system of linear equations
x1 + 2x2 + 3x3 = 1
4x1 + 5x2 + 6x3 = −1
7x1 − 5x2 − 9x3 = 0
c
2019 School of Mathematics and Statistics, UNSW Sydney
136 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Example 4. Write the system of linear equations corresponding to the augmented matrix
1 3 6 7 −2
−2 0 5 −4 3 .
7 0 0 −5 −10
Solution. Because there are 4 columns before the “|”, the system has 4 variables, which we will
call x1 , x2 , x3 and x4 (although other names are acceptable). Because (A|b) has 3 rows, the system
has 3 equations. Reading off the coefficients of row 1, row 2, and row 3 in turn, we obtain the
system
x1 + 3x2 − 6x3 + 7x4 = −2
−2x1 + 5x3 − 4x4 = 3
7x1 − 5x4 = −10
Although it is conventional to omit variables with zero coefficients when writing a system of linear
equations, it is essential in matrix notation that every row contains exactly the same number of
entries and that every column also contains exactly the same number of entries so 0s must not be
omitted from matrices or vectors. ♦
Example 5. Write the system of linear equations corresponding to the matrix equation Ax = b,
where
3 −2 6 7 −8 7
A= , b= .
5 3 −2 −7 4 −3
Solution. The matrix A has 5 columns, so there are 5 variables, which we call x1 , x2 , x3 , x4 , x5 .
Also, A has 2 rows, so there are 2 equations.
Reading off the coefficients from the two rows, we obtain the system
3x1 − 2x2 + 6x3 + 7x4 − 8x5 = 7
5x1 + 3x2 − 2x3 − 7x4 + 4x5 = −3
♦
Example 6. Write down the Ax = b form of the system of linear equations corresponding to the
vector equation
2 5 −4 6 0
x1 −3 + x2 0 + x3 8 + x4 7 = −6 .
4 −1 0 −2 2
Solution. The vector associated with xj is the jth column of the coefficient matrix so as a matrix
equation:
x1
2 5 −4 6 x 2 0
−3 0 8 7 −6 .
x 3 =
4 −1 0 −2 2
x4
♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.3. ELEMENTARY ROW OPERATIONS 137
(1′ ) = (2) −x1 + 3x2 + x3 = 6 New R1 = old R2 −1 3 1 6
(2′ ) = (1) 2x2 + 3x3 = 5 or New R2 = old R1 0 2 3 5 .
(3) 2x1 + 4x2 + 7x3 = 8 R3 2 4 7 8
c
2019 School of Mathematics and Statistics, UNSW Sydney
138 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Proposition 1. The solution of the system (1) is the same as the solution of the system (2).
x1
..
Proof. Let x = . be a solution of system (1). Then we have
xn
a1 x1 + . . . + an xn = b
and (α1 + λa1 )x1 + · · · + (αn + λan )xn = (α1 x1 + · · · + αn xn ) + λ(a1 x1 + · · · + an xn )
= β + λb,
a1 x1 + . . . + an xn = b
and α1 x1 + · · · + αn xn = (α1 + λa1 )x1 + · · · + (αn + λan )xn − λ(a1 x1 + · · · + an xn )
= (β + λb) − λb = β,
In the augmented-matrix notation, the addition of a scalar multiple of one equation to another
is equivalent to addition of a scalar multiple of one complete row to another row.
Example 2. Consider the system
(1) x1 + 2x2 + 3x3 = 5 R1 1 2 3 5
(2) −x1 − 4x2 + x3 = 6 or R2 −1 −4 1 6
(3) 2x1 + 10x2 + 7x3 = 8 R3 2 10 7 8
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.4. SOLVING SYSTEMS OF EQUATIONS 139
We can multiply one equation by a non-zero constant without changing the solutions of the system.
For example the systems of equations corresponding to the two matrices
1 2 3 4 8 1 2 3 4 8
6 9
0 0 5 6 9 and 0 0 1 5 5
3
0 0 0 7 3 0 0 0 1 7
are equivalent.
This should be recorded as
1 2 3 4 8 R2 = 51 R2 1 2 3 4 8
−−−−−−−−−−−→ 6 9
0 0 5 6 9 0 0 1 5 5
.
1
0 0 0 7 3 R 3 = 7 R 3 0 0 0 1 3
7
Row operations should be done one after the other. Although in the above we have done two
row operations at once, there is no difference from performing the operations one after the other.
However, we should not make the following mistake.
1 2 3 5 R2 = R2 + R1 0 −2 4 11
−1 −4 1 6 −−−−−−−−−−−−−−→ 0 −2 4 11
2 10 7 8 R1 = R1 + R2 2 10 7 8
The two systems are not equivalent. If we do want to perform the row operations R2 = R2 + R1
then R1 = R1 + R2 , we should have
1 2 3 5 1 2 3 5 1 0 7 16
−1 −4 1 6 −−R−2−
=R2 +R1 R =R1 +R2
−−−−−→ 0 −2 4 11 −−−1−−−−−−→ 0 −2 4 11 .
2 10 7 8 2 10 7 8 2 10 7 8
c
2019 School of Mathematics and Statistics, UNSW Sydney
140 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
1. In the first stage, known as Gaussian elimination, we use two types of row operation
(interchange of rows and subtracting a multiple of one row from another row) to produce
an equivalent system in a simpler form which is known as row-echelon form. From the
row-echelon form we can tell many things about solutions of the system. In particular, we
can tell whether the system has no solution, a unique solution or infinitely many solutions.
2. If the system does have solutions, the second stage is to find them. It can be carried out by
either of two methods.
(a) We can use further row operations to obtain an even simpler form which is called re-
duced row-echelon form. From this form we can read off the solution(s).
(b) From the row echelon form, we can read off the value (possibly in terms of parameters)
for at least one of the variables. We substitute this value into one of the other equations
and get the value for another variable, and so on. This process, which is called back-
substitution, will be described fully later.
2. the leading entry in a leading row is the first (i.e. leftmost) non-zero entry,
3. a leading column is a column which contains the leading entry for some row.
For example, in the following matrix the 1st row is a leading row with leading entry 5 and the 2nd
row is a non-leading row. The 2nd column is a leading column but the 1st and 3rd columns are
non-leading columns.
0
5 7
0 0 0
1. all leading rows are above all non-leading rows (so any all-zero rows are at the
bottom of the matrix), and
2. in every leading row, the leading entry is further to the right than the leading
entry in any row higher up in the matrix.
The following are examples of augmented matrices which are in row-echelon form. The circled
entries are leading entries. We shall see later that the position of leading entries in a row-echelon
form gives important information about solubility of the system.
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.4. SOLVING SYSTEMS OF EQUATIONS 141
2 3 4 11 −5
0 4 1 0
(1) 0
−3 2
7 ,
(2) 0
2 3 2
1
,
0 0
4 8 0 0 0 0
2
−5
0 2 4 1 0
6 2 3 4
(3)
0 0
7 7
, (4) 0
0
2 3 2 1
,
0 0 0 0 0 0
2 3 4 11
0
−3 2 7
−5
0 4 1 0
(5)
,
(6) 0
2 3 2 0 .
0
0
4 8
0 0 0 0 0
0 0 0 0
The following are examples of augmented matrices which are NOT in row-echelon form.
0
6 3 4
−5
0 4 0
(7) , (8) 0 0 0 0 ,
5 7 0 0
0
2 3 1
5 3 5 −4 6
(9) 0 0
6 −6 7
0 0
7 0 −8.
Example (8) does not satisfy Condition 1 of Definition 2. Examples (7) and (9) do not satisfy
Condition 2.
Warning. The definition of row-echelon form varies from one book to another and the terminology
in Definition 1 is not universally adopted.
The following are examples of augmented matrices which are in reduced row-echelon form.
c
2019 School of Mathematics and Statistics, UNSW Sydney
142 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
1 0 0 1
0
1 2 0
(10) 0
1 0 2 , (11) 0 0 0
1 ,
0 0
1 3 0 0 0 0
1 −1 0 2 0 2
(12) 0 0
1 1 0 −1 .
0 0 0 0
1 6
Later in this section, we shall see that we can easily read the solutions of a system with its
augmented matrix in reduced row-echelon form. The solution set of a system with its augmented
matrix in row-echelon from can be found by a process called back-substitution.
In example (a), row 1 is the pivot row and column 1 is the pivot column. In example (b), row 2 is
the pivot row and column 2 is the pivot column.
Step 2. By a row interchange, swap the pivot row and the top row if necessary.
The rule is that the first pivot row of the augmented matrix must finish as the first row of
(U |y). You can achieve this by interchanging row 1 and the pivot row.
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.4. SOLVING SYSTEMS OF EQUATIONS 143
In example (a), the pivot row is already row 1, so no row interchange is needed. In example
(b), the pivot row is row 2, so rows 1 and 2 of the augmented matrix must be interchanged. This
is shown as
0 0 0 2 3 −1 0
3 −3 3 3 −6
0
3 −3 3 3 −6 R1 ↔ R2 0 0 0 2 3 −1
b) −−−−−−−−−−→ .
0 0 0 1 3 −2 0 0 0 1 3 −2
0 6 −6 6 3 −9 0 6 −6 6 3 −9
Step 3. Eliminate (i.e., reduce to 0) all entries in the pivot column below the pivot
element.
We can do this by adding suitable multiples of the pivot row to the lower rows. After this
process the pivot column is in the correct form for the final row-echelon matrix (U |y).
In example (a), we can use the row operations R2 = R2 + (−2) R1 and R3 = R3 + (−1) R1 to
reduce the pivot column to the required form. In recording the row operations, we can simply write
R2 = R2 − 2R1 and R3 = R3 − R1 .
1 2 1 3
R2 = R2 − 2R1
1 2 1 3
a) 2 3 1 3
−−−−−−−−−−−−−−→ 0 −1 −1 −3 .
R3 = R3 − R 1
1 3 3 2 0 1 2 −1
In example (b), the row operation R4 = R4 − 2 R1 reduces the pivot column to the required
form. This gives
0
3 −3 3 3 −6
0
3 −3 3 3 −6
0 0 0 2 3 −1 R4 = R4 − 2R1 0 0 0 2 3 −1
b) −−−−−−−−−−−−−−−→ .
0 0 0 1 3 −2 0 0 0 1 3 −2
0 6 −6 6 3 −9 0 0 0 0 −3 3
Step 4. Repeat steps 1 to 3 on the submatrix of rows and columns strictly to the right
of and below the pivot element and stop when the augmented matrix is in row-echelon
form.
Note that the top row in step 2 here should mean the top row of the submatrix. In example
(a), the required operation is
1 2 1 3
R3 = R3 + R2
1 2 1 3
0 −1
−1 −3 −−−−−−−−−−−−−−→ 0 −1
−1 −3 .
0 1 2 −1 0 0 1 −4
Then we have reduced the matrix to row-echelon from
1 2 1 3
0 −1
−1 −3 .
0 0
1 −4
c
2019 School of Mathematics and Statistics, UNSW Sydney
144 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
In both examples we have now reached a matrix which is in row-echelon form, so we have
completed the process of Gaussian elimination for these examples.
Note.
1. If you are doing Gaussian elimination without the aid of a computer, you don’t have to stick
rigidly to the pivot selection rule which we stated above. For example, the value 1 is very
convenient as a pivot entry, so when the leftmost non-zero column has a 1 in it you may find
it best to use this 1 as your pivot entry in preference to some other non-zero entry which is
above the 1.
For example, if you have the augmented matrix
13 27 174
1 2 13
you should do
13 27 174 R1 ↔ R2 1 2 13
−−−−−−−−−−→
1 2 13 13 27 174
R2 = R2 − 13R1 1 2 13 R1 = R1 − 2R2 1 0 3
−−−−−−−−−−−−−−−−→ −−−−−−−−−−−−−−−→
0 1 5 0 1 5
rather than applying the rule stated above, which would give
1
13 27 174 R2 = R2 − 13 R1 13 27 174
−−−−−−−−−−−−−−−−→ 1 5 etc.
1 2 13 0 − 13 − 13
Ri ↔ Rj ,
Ri = λRi where λ 6= 0,
Ri = Ri + λRj ,
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.4. SOLVING SYSTEMS OF EQUATIONS 145
and they are supposed to be done one after the other. Be careful to avoid doing simultaneously
two operations like
R1 = R1 − 2R2 and R2 = R1 − 2R2 .
See page 139. If all the entries in Ri are integers and have a common factor λ, we should use
Ri = λ1 Ri to produce integers with smaller magnitude.
x1 + 0x2 + 0x3 = 3
0x1 + x2 + 0x3 = −1
0x1 + 0x2 + x3 = 2,
Notice that in the original row-echelon form, every column of the coefficient matrix (the
part of the augmented matrix to the left of the vertical bar) is a leading column and that the
system of equations has a unique solution.
Example 2. In this case we can get to reduced row-echelon form just by multiplying each row by
a suitable constant — R1 = − 15 R1 , R2 = 21 R2 and R3 = 12 R3 . The reduced row-echelon form is
c
2019 School of Mathematics and Statistics, UNSW Sydney
146 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
then
1 0 − 45 − 51 0
3 1
0 1 2 1 2
.
0 0 0 0 1
Since the third row of the reduced row-echelon form represents the equation
(0 0 0 · · · 0 | α), where α 6= 0,
then the corresponding system of equations has no solution. Note that if the right-hand-side
column is a leading column, then the system of equations has no solution, i.e. inconsistent.
Example 3. The transformation to reduced row-echelon form goes like this.
6 2 3 4 −− R1 = R1 − 3R2 6 2 0 1
−−−−−−−−−−−→
R 1 =
−−−−−−−
1
6 R 1
−−→
1 1
3 0 1
6 .
0 0 1 4 0 0 1 1 0 0
1 1
x1 + 31 x2 = 16 , x3 = 1.
The second equations tells us that x3 must equal 1. For the first equation, we need to set x2 as a
parameter, say λ. This means that the system must have infinitely many solutions. In fact x will
be a solution if and only if
1
x1 = 6 − 13 λ, x2 = λ, x3 = 1.
This is a parametric vector form of a line in R3 through the point ( 16 , 0, 1) parallel to the vector
1
−3
1 . ♦
0
A variable xi is a leading variable if the ith column of the row-echelon matrix is a leading
column. It is a non-leading variable if the ith column of the matrix is a non-leading column. Notice
that the leading variables in the echelon form are x1 and x3 while x2 is non-leading, so this example
illustrates a general rule which says that non-leading variables can be chosen arbitrarily by
setting parameter, and then the leading variables can be written in terms of the parameters.
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.4. SOLVING SYSTEMS OF EQUATIONS 147
Columns 2, 4 and 5 are non-leading columns, so the variables x2 , x4 and x5 are non-leading
variables and in terms of parameters we have x2 = λ1 , x4 = λ2 and x5 = λ3 . We can then read off
the values of the leading variables as
1
x1 = 5 + 51 λ2 − 51 λ3 , x3 = 1
2 − 23 λ2 − λ3 .
This set of solutions can be expressed in vector form as
1 1 1
x1 5 0 5 −5
x2 0 1 0 0
1 3
x=
x3 = 2 + λ1 0 + λ2 − 2 + λ3 −1 for λ1 , λ2 , λ3 ∈ R.
x4 0 0 1 0
x5 0 0 0 1
♦
Examples 3 and 4 illustrate that the parameters in the solution are the values of the
non-leading variables, so the number of parameters in the solution equals the number
of non-leading columns in the row-echelon form of the coefficient matrix.
In general, if a system is consistent and there are non-leading columns in a row-echelon form of
the system, then the system has infinitely many solutions.
Example 5. In this example, row 4 is all zeros and so is superfluous. The remaining rows are the
same as in example 1, so the solution is identical to the solution in example 1. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
148 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
We end this subsection by summarising the rule for getting solutions from a reduced echelon
form of the system:
Assign an arbitrary parameter as the value for each non-leading variable. Then read off expressions
for the leading variables in terms of the arbitrary parameters which you have introduced.
4.4.4 Back-substitution
When solving small systems by hand, you may prefer to use this procedure as an alternative to the
transformation of a row-echelon form into reduced row-echelon form. Instead of doing further row
operations on the augmented matrix, we go to the equations represented by the row-echelon form
and proceed as follows.
Assign an arbitrary parameter value to each non-leading variable. Then read off from the last non-
trivial equation an expression for the last leading variable in terms of your arbitrary parameters.
Substitute this expression back into the second last equation to get an expression for the second last
leading variable, and so on.
Example 7. We will redo example 4 of the last subsection, using back-substitution instead of
transformation to reduced echelon form. The row-echelon form is
−5 0 2 4 1 0
0 0
2 3 2 1 .
0 0 0 0 0 0
which gives
1 3
x3 = − λ2 − λ3 .
2 2
Substituting this back into the equation represented by the first row gives
1 3
−5x1 + 2 − λ2 − λ3 + 4λ2 + λ3 = 0.
2 2
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.5. DEDUCING SOLUBILITY FROM ROW-ECHELON FORM 149
Proposition 1. If the augmented matrix for a system of linear equations can be transformed into
an equivalent row-echelon form (U |y) then:
1. The system has no solution if and only if the right hand column y is a leading column.
2. If the right hand column y is a NOT a leading column then the system has:
It often happens in practice that you want to consider the system Ax = b for one fixed coefficient
matrix A but many different vectors b. For this reason it is useful to know the answer to the
following question: if we know a row-echelon form U for the coefficient matrix A (not the augmented
matrix (A|b)), what can we say about solubility of the system Ax = b for an arbitrary right hand
side b?
If U has an all-zero row then it will obviously be possible to find at least one b ∈ Rn such that
the row-echelon form (U |y) for (A|b) has a row of the form
0 0 · · · 0 | α , α 6= 0
and this represents an equation which has no solution. On the other hand, if U has no non-leading
row (see page 140) then no b can give rise to an impossible row of the above type and we can
always get a solution by back-substitution. In this case the number of solutions will be infinite if
and only if there are non-leading columns in U .
The following proposition sums up what we can say about solubility for arbitrary b.
a) at least one solution for each b in Rm if and only if U has no non-leading rows,
b) at most one solution for each b in Rm if and only if U has no non-leading columns,
c) exactly one solution for each b in Rm if and only if U has no non-leading rows and no
non-leading columns.
When U has neither non-leading row nor non-leading column, Ax = b always has a unique
solution. Otherwise, whether the matrix equation has solution depends on the vector b. We shall
study examples on finding a condition on b for a solution to exist and we shall discuss how to find
a specific formula for a solution x expressed in terms of an arbitrary right hand side b in the next
section.
c
2019 School of Mathematics and Statistics, UNSW Sydney
150 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
1 3
− x2 = − b1 + b2 or x2 = 3b1 − 2b2 .
2 2
Substitute it into the equation represented by the first row,
♦
Note. We could write the system of equations as
2x1 + 3x2 = 1b1 + 0b2
3x1 + 4x2 = 0b1 + 1b2
Provided the interpretation of the columns is clearly understood the row reduction can be rewritten
as
2 3 1 0 R2 = R2 − 32 R1 2 3 1 0
−−−−−−−−−−−−−→ .
3 4 0 1 0 − 21 − 32 1
Then we can proceed to reduce the matrix to reduced row-echelon from or apply back-substitution
as before.
In the above example, the system always has a unique solution for any b. The system of
equations in the next example does not always have solution. A condition on b has to be satisfied
for a solution to exist.
b1
Example 2. Find a condition for b = so that the system of equations
b2
2x1 + 3x2 = b1
4x1 + 6x2 = b2
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.7. GENERAL PROPERTIES OF THE SOLUTION OF Ax = b 151
x = x p + λ 1 v 1 + λ 2 v 2 + · · · + λk v k ,
c
2019 School of Mathematics and Statistics, UNSW Sydney
152 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
where v1 , · · · , vk are non-zero vectors and where the parameters λ1 , . . . , λk are scalars. It is not
difficult to see that this is generally true for all systems of linear equations with infinitely many
solutions.
Furthermore, we shall prove that the vector xp is a solution of the system Ax = b, and the
vectors vi which go with the parameters in the solution are solutions of the homogeneous equation
Ax = 0. It does not depend on how we reduce the matrix to an echelon form.
We begin with some important propositions about solutions of Ax = 0.
Proposition 1. Ax = 0 has x = 0 as a solution.
Proof. Since Ax = 0 represents the system of equations
a11 x1 + a12 x2 + ··· + a1n xn = 0
a21 x1 + a22 x2 + ··· + a2n xn = 0
.. .. .. .. ,
. . . .
am1 x1 + am2 x2 + · · · + amn xn = 0
so x1 = x2 = · · · = xn = 0 satisfies all the equations.
Proposition 2. If v and w are solutions of Ax = 0 then so are v + w and λv for any scalar λ.
v1 w1
.. ..
Proof. Since v = . and w = . are solutions, we have
vn wn
a11 v1 + a12 v2 + ··· + a1n vn = 0
a21 v1 + a22 v2 + ··· + a2n vn = 0
.. .. .. .. , (1)
. . . .
am1 v1 + am2 v2 + · · · + amn vn = 0
a11 w1 + a12 w2 + ··· + a1n wn = 0
a21 w1 + a22 w2 + ··· + a2n wn = 0
.. .. .. .. . (2)
. . . .
am1 w1 + am2 w2 + · · · + amn wn = 0
If we add the ith equations of (1) and (2) for 1 6 i 6 m, we shall get
a11 (v1 + w1 ) + a12 (v2 + w2 ) + ··· + a1n (vn + wn ) = 0
a21 (v1 + w1 ) + a22 (v2 + w2 ) + ··· + a2n (vn + wn ) = 0
.. .. .. .. .
. . . .
am1 (v1 + w1 ) + am2 (v2 + w2 ) + · · · + amn (vn + wn ) = 0
Hence v + w is a solution of Ax = 0.
If we multiply both sides of every equation of (1) by λ, then
a11 (λv1 ) + a12 (λv2 ) + · · · + a1n (λvn ) = 0
a21 (λv1 ) + a22 (λv2 ) + · · · + a2n (λvn ) = 0
.. .. .. .. .
. . . .
am1 (λv1 ) + am2 (λv2 ) + · · · + amn (λvn ) = 0
Hence λv is also a solution.
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.7. GENERAL PROPERTIES OF THE SOLUTION OF Ax = b 153
A further important proposition about homogeneous systems can be obtained by induction from
the last proposition.
Proposition 3. Let v1 , . . . , vk be solutions of Ax = 0 for 1 6 j 6 k. Then λ1 v1 + · · · + λk vk is
also a solution of Ax = 0 for all λ1 , . . . , λk ∈ R.
We now turn to the properties of the solutions of Ax = b.
Proposition 4. If v and w are solutions of Ax = b then v − w is a solution of Ax = 0.
v1 w1
.. ..
Proof. Since v = . and w = . are solutions of Ax = b, we have
vn wn
a11 v1 + a12 v2 + ··· + a1n vn = b1
a21 v1 + a22 v2 + ··· + a2n vn = b2
.. .. .. .. , (1)
. . . .
am1 v1 + am2 v2 + · · · + amn vn = bn
a11 w1 + a12 w2 + ··· + a1n wn = b1
a21 w1 + a22 w2 + ··· + a2n wn = b2
.. .. .. .. . (2)
. . . .
am1 w1 + am2 w2 + · · · + amn wn = bn
If we subtract the ith equation of (2) from the ith equation of (1) for 1 6 i 6 m, we shall get
a11 (v1 − w1 ) + a12 (v2 − w2 ) + ··· + a1n (vn − wn ) = 0
a21 (v1 − w1 ) + a22 (v2 − w2 ) + ··· + a2n (vn − wn ) = 0
.. .. .. .. .
. . . .
am1 (v1 − w1 ) + am2 (v2 − w2 ) + · · · + amn (vn − wn ) = 0
Hence v − w is a solution of Ax = 0.
c
2019 School of Mathematics and Statistics, UNSW Sydney
154 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Note. The form of solutions in part 2 of Proposition 5 raises the question of what is the minimum
number of vectors v1 , · · · , vk required to yield all the solutions of Ax = b. We will return to this
question in Chapter 6 when we discuss the ideas of “spanning sets” and “linear independence”. For
the present, it is sufficient to note that the solution method we have used in this chapter does find
all solutions of Ax = b.
Example 1. In Section 4.4.2 we found a row-echelon form for the augmented matrix of Ax = b,
where
x1
0 0 0 2 3 x2 −1
0 3 −3 3 3
A= , x = x3 and b = −6
0 0 0 1 3 −2
x4
0 6 −6 6 3 −9
x5
as
0 3 −3 3 3 −6
0 0 0 2 3 −1
0 0 0 0 3 −3 .
2 2
0 0 0 0 0 0
By back-substitution, x5 = −1, x4 = 1. By setting x1 = λ1 and x3 = λ2 , we also have x2 = λ2 − 2.
Hence the solutions are
λ1 0 1 0
λ2 − 2 −2 0 1
x=
λ2 = 0 + λ1 0 + λ2 1 for λ1 , λ2 ∈ R.
1 1 0 0
−1 −1 0 0
The first vector is a particular solution (corresponding to λ1 = λ2 = 0) of Ax = b and the vectors
which go with the parameters are solutions of the homogeneous equation Ax = 0.
As an exercise, can you write down the two systems of 4 equations in 5 variables corresponding
to Ax = b and Ax = 0 and verify that the first vector is a solution to the non-homogeneous system
and the two other vectors are solutions to the homogeneous system? ♦
4.8 Applications
Linear equations appear in many applications. In this section we first present some geometric
applications and then some applications in some other fields.
4.8.1 Geometry
In Section 4.1, we have shown some simple geometric applications of linear equations to lines in R2
and planes in R3 . In this section we show how some common problems involving lines and planes
in Rn can be solved using linear equations.
3 1 0
0 −2 4
Example 1. Does
5 belong to span 3 , 1 ?
6 2 2
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.8. APPLICATIONS 155
3 1 0
0 −2 4
Solution. The vector
5 ∈ span 3 , 1 if and only if
6 2 2
3 1 0
0 −2 4
= λ1
5 3 + λ2 1 for some λ1 , λ2 ∈ R.
6 2 2
2
Example 2. Is v = 4 parallel to the plane
6
1 −1 3
x = 2 + λ1 3 + λ2 1 ?
3 2 4
2
Solution. The vector v = 4 is parallel to the plane if and only if it is a linear combination
6
−1 3
of the two vectors 3 and 1 . Hence, v is parallel to the plane if and only if there are real
2 4
c
2019 School of Mathematics and Statistics, UNSW Sydney
156 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Since
the right hand column is non-leading, this system has a solution for λ1 and λ2 , and hence
2
4 is parallel to the plane. ♦
6
1
2
Example 3. Find the intersection of the line through (3, 2, 1, 4) parallel to
−3 and the plane
6
2 −1
1 3
through (3, 1, −4, 7) parallel to
4 and 0 .
5 6
Solution. The equation of the line is
3 1
2 2
x=
1 + λ −3 for λ ∈ R,
4 6
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.8. APPLICATIONS 157
The system of equations has no solution, so the given line and plane in R4 do not
intersect. ♦
The oil company has to decide how many hours per day to run each refinery in order to produce
a required amount of product each day.
Do the following:
c
2019 School of Mathematics and Statistics, UNSW Sydney
158 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Solution. Let
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.8. APPLICATIONS 159
is equivalent to
10000 2000 5000 610000
0 1000 0 15000
(U |y) =
0 0 −50 −1100
0 0 0 600
which has no solution, so production levels can not be met. ♦
The model used here is too simple to reflect accurately what is done in practice. In practice
the problem is generalised in two ways. First, the equations are replaced by inequalities; i.e., the
problem is changed from one of having to produce exactly the required amount of product to one of
having to produce at least the required amount of product. Secondly, the problem is changed to one
of having to minimise the cost of production subject to the constraints of having to produce the
required amount of product. This generalised problem is called a linear programming problem.
Linear programming is very important in economic, financial and manufacturing applications and
so on where problems involving thousands of variables and inequalities are routinely solved, and
result in savings of millions of dollars.
4.8.3 Economics
A simplified example of an application of linear equations to economics is as follows.
Example 5. The island of Wotsit-Matta has a simple economy in which the only goods produced
are wheat, iron and pigs. The Wotsit-Mattas have worked out that the following amounts of
wheat, iron and pigs are required on 31 December of each year for each unit of wheat, iron and
pigs produced in the following year:
That is, to produce 1 tonne of wheat in a particular year we need to have available on 31
December of the previous year 0.5333 tonnes of wheat, 0.02557 tonnes of iron and 0.0400 hundred
pigs, and similarly for iron and pigs.
Assuming that all of the wheat, iron and pigs available at 31 December of one year are used to
produce wheat, iron and pigs in the following year, find the amount of wheat, iron and pigs at 31
December, 1988 and then at 31 December, 1989 if
(a) the Wotsit-Mattas have 180 tonnes of wheat, 8.401 tonnes of iron and 2400 pigs available at
31 December, 1987, and
(b) the Wotsit-Mattas have 180 tonnes of wheat, 8 tonnes of iron and 2400 pigs available at 31
December, 1987.
c
2019 School of Mathematics and Statistics, UNSW Sydney
160 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
Solution. We first set up a linear equations model for the Wotsit-Matta economy. Note that the
unknowns are the amounts of wheat, iron and pigs produced in a year, and the right-hand sides
are the amounts available from the previous year.
We let
x1 = number of tonnes of wheat produced in a year
x2 = number of tonnes of iron produced in a year
x3 = number of hundreds of pigs produced in a year
and let
b1 = number of tonnes of wheat available from the previous year
b2 = number of tonnes of iron available from the previous year
b3 = number of hundreds of pigs available from the previous year
We can now solve for the production in years 1988 and 1989.
180
CASE 1.(a). On 31 December, 1987, b = 8.401 . On solving for the production in 1988 we
24
180
obtain x = 8.403 , that is, 180 tonnes wheat, 8.403 tonnes iron and 2400 pigs.
24
This 1988 production is then
usedas the right-hand side b in the equations forthe 1989
180 179.9
production. On solving Ax = 8.403 as before, we obtain the solution x = 8.425 , and so
24 23.98
the 1989 production is 179.9 tonnes wheat, 8.425 tonnes iron and 2398 pigs.
180
CASE 2.(b). On 31 December, 1987, b = 8 . We find that the solution for 1988 production
24
is 200.1 tonnes wheat, 4.658 tonnes iron and 2667 pigs.
On using this 1988 production as the right-hand side for the 1989 production, we obtain the
1989 production as 432.2 tonnes wheat, −34.37 tonnes iron and 5787 pigs.
♦
Note. The negative result for iron in case (b) means that the Wotsit-Matta economy is in deep
trouble. Possible actions that the Wotsit-Mattas might take include importing iron from Australia
(if they can find the foreign exchange) or alternatively, having a great party and eating sufficient
wheat and pigs to stop the overheating in these sections of their economy. A complete mathematical
explanation of the marked difference in behaviour between cases (a) and (b) requires a knowledge
of eigenvalues and eigenvectors (which will be studied in MATH1231).
c
2019 School of Mathematics and Statistics, UNSW Sydney
4.9. MATRIX REDUCTION AND MAPLE 161
c
2019 School of Mathematics and Statistics, UNSW Sydney
162 CHAPTER 4. LINEAR EQUATIONS AND MATRICES
BackwardSubstitute(%);
Experimenting with this may help motivate the reduced row-echelon form. You should also exper-
iment with:
BackwardSubstitute(Ab);
to see what can go wrong if you leave out the Gaussian elimination step.
Rows 1 and 3 of the matrix C may be swapped with the command:
RowOperation(C,[1,3]);
You may replace row 4 by row 4 plus 72 times row 3 in the matrix A to row 4 with
RowOperation(A,[4,3],7/2);
You may multiply row 4 of the matrix A by 72 by the command:
RowOperation(A,4,7/2);
You should now practise reducing a ‘random’ matrix A by hand and comparing your answer with
that obtained from GaussianElimination and ReducedRowEchelonForm. Other useful commands
may be found by typing in:
?Pivot
?LinearSolve
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 4. LINEAR EQUATIONS AND MATRICES 163
1. [R] Find the solution set of each of the following linear equation.
a) 2x1 − 5 = 0 as an equation of one variable, then as an equation in two variables, and
then three variables.
b) x1 + 2x2 = 4 as an equation of two variables, then three variables.
c) 2x1 − 3x2 + x3 = 2 as an equation of three variables.
2. [R][V] Determine algebraically whether the following systems of equations have a unique solution,
no solution, or an infinite number of solutions. Draw graphs to illustrate your answers.
a) 3x1 + 2x2 = 6 b) 3x1 + 2x2 = 6
9x1 + 6x2 = 36 9x1 + 4x2 = 36
c) x1 − 5x2 = 5
6x1 − 30x2 = 30
3. [H] Find conditions on the coefficients a11 , a12 , a21 , a22 , b1 , b2 so that the system of equations
a11 x1 + a12 x2 = b1
a21 x1 + a22 x2 = b2
has a) a unique solution, b) no solution, and c) an infinite number of solutions.
For simplicity, assume a11 6= 0.
4. [X] Repeat the previous question with no simplifying assumptions. That is, find general
conditions which apply for all possible values of the coefficients.
5. [R] Find and geometrically describe the solutions for the following systems of linear equations.
a) x1 + 2x2 + 3x3 = 5 b) 4x1 + 5x2 − 2x3 = 16
2x1 + 5x2 + 8x3 = 12 8x1 + 10x2 − 4x3 = 20
c) 4x1 + 5x2 − 2x3 = 16
8x1 + 10x2 − 4x3 = 32
6. [X] Prove algebraically that two distinct planes in R3 either intersect in a line or are parallel
with no points in common. Use a linear equation in three unknowns to represent a plane
in R3 .
7. [R] Show that x1 = 2 − 2λ, x2 = λ, x3 = 3 + 2λ, where λ is any real number, satisfy the
system of equations
x1 + 4x2 − x3 = −1
2x1 + 4x2 = 4
6x2 − 3x3 = −9
c
2019 School of Mathematics and Statistics, UNSW Sydney
164 PROBLEMS FOR CHAPTER 4
8. [R][V] Write each of the following systems of equations in vector form, as a matrix equation
Ax = b, and in the augmented matrix (A|b) form.
9. [R] Write the system of equations, the matrix equation and the augmented matrix form cor-
responding to the vector equation
1 −3 0 10
0 6 6 −2
x1
−6 + x2 −1 + x3 −4 = 0 .
7 9 11 5
10. [R] For each of the following matrices, find the appropriate elementary row operations to
describe the transformation from one matrix to the next. Also continue the row reduction
until the matrix is in row echelon form.
1 4 2 3 1 4 2 3
a) 2 6 3 0 → 0 −2 −1 −6 ,
4 −2 4 4 0 −18 −4 −8
3 4 1 3 1 −4 1 1
b) 2 8 0 2 → 1 4 0 1 .
0 8 3 0 0 8 3 0
11. [M] Write down the output when the Maple command RowOperation(A,[2,1],3); is applied
to the matrix
2 4 1 2
A = 3 2 4 1 .
1 3 1 3
12. [R] For each of the following augmented matrices do the following. Determine whether the ma-
trix is in row-echelon form as defined in Section 4.4.1. If the matrix is in row-echelon form,
identify the leading elements, leading rows, leading columns, and non-leading columns.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 4. LINEAR EQUATIONS AND MATRICES 165
3 2 1 10 3 2 1 10
a) 0 4 2 8 , b) 3 2 1 10 , c) 4 0 2 8 ,
0 0 −7 14 0 0 −7 14
3 2 1 10
3 2 1 10 0 3 1 6 0 4 2 8
d) , e) , f)
0 0 −7
,
0 4 2 8 0 0 1 5 14
0 0 0 0
3 2 1 10 3 2 1 10
0 4 2 8 0 4 2 8
g)
0
, h) .
0 −7 14 0 0 0 0
0 0 0 6 0 0 0 6
13. [R] Find the solutions to the following systems of equations. If possible give a geometric
interpretation of the solution.
a) 3x1 + 2x2 + x3 = 10
4x2 + 2x3 = 8
− 7x3 = 14
b) 3x1 + 2x2 + x3 + x4 = 10
4x2 + 2x3 − 4x4 = 8
− 7x3 + 14x4 = 14
14. [R][V] For each of the following systems of equations, do the following:
i) Write down the corresponding augmented matrix.
ii) Use Gaussian elimination to transform the augmented matrix into row-echelon form.
iii) Solve each system of equations writing your answer in vector form.
a) x1 − 2x2 = 5 b) x1 − 2x2 − 3x3 = 3
3x1 + x2 = 8 2x1 + 4x2 + 10x3 = 14
g) x1 + 2x2 − x3 + x4 = 4
x2 − x3 + x4 = 1
3x1 + 2x2 − 2x4 = 3
5x1 + 3x2 − x4 = 9
h) x1 + 2x2 − x3 + x4 = 4
x2 − x3 + x4 = 1
3x1 + 2x2 − 2x4 = 3
c
2019 School of Mathematics and Statistics, UNSW Sydney
166 PROBLEMS FOR CHAPTER 4
15. [R] For each of the following augmented matrices, find a reduced row-echelon form. Then write
down all solutions of the corresponding system of equations and try to give a geometric
interpretation of the solutions.
2 4 1 4 1 2 3 4 1
a) 0 1 2 −2 , b) 0 −1 5 6 2 .
0 0 −1 2 0 0 1 7 3
16. [R][V] For each of the following augmented matrices, without solving, decide whether or not
the corresponding system of linear equations has a unique solution, no solution or infinitely
many solutions.
3 2 1 10
3 2 1 10 0 4 2 8
a) 0 4 2 8 , b) 0 0 −7 14
, c) 3 2 1 10 ,
0 0 −7 14
0 0 0 6
3 2 1 10
3 2 1 10 0 4 2 8
d) , e) 0 0 −7 0 .
0 4 2 8
0 0 0 0
17. [H][V] Determine which values of k, if any, will give a) a unique solution b) no solution
c) infinitely many solutions to the system of equations
x + y + kz = 2
3x + 4y + 2z = k
2x + 3y − z = 1.
x + 2y + λz = 1
−x + λy − z = 0
λx − 4y + λz = −1
have a) no solutions, b) infinitely many solutions, c) a unique solution?
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 4. LINEAR EQUATIONS AND MATRICES 167
20. [H] You are an auditor for a company whose four executives make regular business trips on
four routes and you suspect that at least one of the executives has been overstating her
expenses. You don’t know how much it costs to travel each route, but you know that it is
the same for all the executives. You know the number of trips each executive made on each
route in a certain period and you know the total expenses claimed by each executive for
this period. If the numbers of trips are as shown in the table below, do you have sufficient
information to be sure that someone is cheating? State your reasoning clearly.
Route
1 2 3 4
Executive A 0 1 1 2
Executive B 1 2 0 1
Executive C 3 4 0 1
Executive D 2 1 3 3
21. [H][V] P, Q, R and S are four cities connected by highways which are labelled as shown in the
diagram.
R
a b
P e S
c d
Q
A hire car operator in P makes a note of the number of kilometres travelled by five
customers who made trips starting and ending at P . He knows that the routes travelled
by the five customers were as follows: abdc abdea cddc cdbec aedbec
Can he determine the length of each of the five highways? State your reasoning clearly.
22. [R] For each of the following systems of linear equations, find x1 , x2 and x3 in terms of b1 , b2
and b3 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
168 PROBLEMS FOR CHAPTER 4
24. [R] For the following systems, find conditions on the right-hand-side vector b which ensure
that the system has a solution.
7 −2
25. [R] Show that x = 2 + λ 0 , λ ∈ R are the solutions of
0 1
x1 − 2x2 + 2x3 = 3
2x1 − 6x2 + 4x3 = 2
−2x1 + 4x2 − 4x3 = −6
−2
and that x = λ 0 , λ ∈ R are the solutions of the corresponding homogeneous system
1
x1 − 2x2 + 2x3 = 0
2x1 − 6x2 + 4x3 = 0
−2x1 + 4x2 − 4x3 = 0
1 −1 2
27. [R] Is the vector 3 in span 3 , 1 ?
2 4 3
1 3 4
1 −1 −2
28. [R] Is the vector
4 in span 4 , 4 ?
12 6 3
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 4. LINEAR EQUATIONS AND MATRICES 169
3 1 2
1 0 −1
29. [R] Can
−2 be expressed as a linear combination of −3 and 5 ?
4 7 6
2 1 12 3
30. [R] Do the lines x = 1 + λ1 3 and x = 15 + λ2 1 intersect?
3 2 7 −2
5 2 1 3
31. [R] Is the vector 7 parallel to the plane x = 1 + λ1 2 + λ2 5 for λ1 , λ2 ∈ R?
−1 3 1 1
x−1 y z+1
32. [H] Show that the line = = is parallel to the plane
2 3 −1
1 0
x = λ1 1 + λ2 1 , λ1 , λ2 ∈ R.
0 −1
0 2
33. [H] Find the intersection (if any) of the line x = 18 + µ −3 for µ ∈ R and the plane
1 1
1 1 3
x = 0 + λ1 4 + λ2 1 for λ1 , λ2 ∈ R.
4 1 −2
34. [R] Find the intersection (if any) of the planes 8x1 + 8x2 + x3 = 35 and
6 −2 1
x = −2 + λ1 1 + λ2 1 for λ1 , λ2 ∈ R.
3 3 −1
and
2 3 −1
−4 −1 4
x=
1 + µ1 2 + µ2 2 for µ1 , µ2 ∈ R
3 4 6
parallel?
c
2019 School of Mathematics and Statistics, UNSW Sydney
170 PROBLEMS FOR CHAPTER 4
x + 3y + 2z = 5
2x + y− z= 2
7x + 11y + 4z = 13
x+ y− z =1
2x − 4y + 2z = 2
3x − 3y + z = 3
a) Use Gaussian elimination and back–substitution to find the solution(s), if any, of the
above equations.
b) Use your result in part a) to decide whether the three planes represented by the
equations are parallel, intersecting in a straight line, intersecting at a point or have
some other configuration.
38. [R] Find a polynomial p(x) of degree 2 satisfying p(1) = 5, p(2) = 7, p(3) = 13.
39. [R] The total of the ages of my brother, my sister and myself is 140 years. I am seven times
the difference between their ages (my sister is older than my brother) and in seven years
I will be half their combined ages now. How old are we?
40. [R] In a trip to Asia a traveller spent $90 a day for hotels in Bangkok, $60 a day in Singapore
and $60 a day in Kuala Lumpur. For food the traveller spent $60 a day in Bangkok, $90
a day in Singapore, and $60 a day in Kuala Lumpur. In addition the traveller spent $30
a day in other expenses in each city. The traveller’s diary shows that the total hotel bill
was $1020, total food bill was $960, and total other expenses were $420. Find the number
of days the traveller spent in each city, or show that the diary must be wrong.
41. [R] A dietician is planning a meal consisting of three foods. A serving of the first food contains
5 units of protein, 2 units of carbohydrates and 3 units of iron. A serving of the second
food contains 10 units of protein, 3 units of carbohydrates and 6 units of iron. A serving
of the third food contains 15 units of protein, 2 units of carbohydrates and 1 unit of iron.
How many servings of each food should be used to create a meal containing 55 units of
protein, 13 units of carbohydrates and 17 units of iron?
42. [X] A farmer owns a 12-hectare farm on which he grows wheat, oats and barley. Each hectare
of cereal crop planted has certain requirements for labour, fertiliser and irrigation water
as shown in the following table.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 4. LINEAR EQUATIONS AND MATRICES 171
43. [X] In this problem we shall calculate the area of a spherical triangle. Consider the surface
of a sphere of unit radius of area 4π. A great circle on a sphere is the intersection of
that sphere with a plane through the centre. If two great circles meet at antipodal points
P, P ′ let the angle θ between them be the angle 0 < θ < π between the tangents to the
two circles at P . (π − θ is also the angle between the two great circles). Finally define a
spherical triangle to be the region bounded by 3 great circles meeting A, B, C with angles
α, β, γ.
a) The areas bounded by 2 great circles are called lunes. Show their areas are 2θ, 2θ,
2(π − θ), 2(π − θ).
b) Show the surface of the sphere is divided by a spherical triangle into 8 regions equal
in area in pairs.
c) Use parts a) and b) to set up a simple system of 4 linear equations in the 4 areas.
d) Hence show area ABC = α + β + γ − π.
44. [M] The Maple session below (for which the package LinearAlgebra has been loaded) calcu-
lates the intersection of 3 planes Π1 , Π2 and Π3 .
> A:=<<1,3,2>|<2,6,4>|<-1,-1,-1>>;
c
2019 School of Mathematics and Statistics, UNSW Sydney
172 PROBLEMS FOR CHAPTER 4
1 2 −1
A := 3 6 −1
2 4 −1
> b:=<2,12,7>;
2
b := 12
7
> LinearSolve(A,b);
5 − 2 t2
t2
3
> A:=<<1,3,4,7>|<-2,6,2,-8>|<1,8,7,6>|<a,b,c,d>>;
1 −2 1 a
3 6 8 b
A :=
4
2 7 c
7 −8 6 d
> GaussianElimination(A);
1 −2 1 a
0 12 5 b − 3a
7 3a 5b
0 0 − c− −
6 2 6
0 0 0 d − a + 2b − 3c
>
a) The above is a Maple session designed to calculate where 4 planes in R3 meet. What
are the equations of the planes?
b) What are the condition(s) on a, b, c, d for the planes to meet at a point?
c) If a = 1, b = 2, c = 3 and the planes meet, where do they meet?
c
2019 School of Mathematics and Statistics, UNSW Sydney
173
Chapter 5
MATRICES
“It seems very pretty,” she said when she had finished it,
“but it’s RATHER hard to understand!. . .
. . . Somehow it seems to fill my head with ideas
— only I don’t exactly know what they are!”
Lewis Carroll, Through the Looking Glass.
In Chapter 4, matrices are treated as devices to help in writing systems of linear equations.
They are also important in many other areas of mathematics and their applications.
In this chapter, we shall define the addition of two matrices, the multiplication of two matrices
and the multiplication of a matrix by a scalar. Under these operations of addition and scalar
multiplication matrices behave much like vectors and indeed like real numbers. For instance, both
matrices and real numbers obey associative laws and distributive laws. Yet, there are differences
between matrices and real numbers. Multiplication of real numbers is commutative, but, as we
shall see, matrix multiplication is not. Every non-zero real number has an multiplicative inverse
which is the reciprocal of that number. However, not all matrices have inverses. We shall study
the matrix arithmetic and algebra and an important function on the set of square matrices—the
determinant function.
In this section we describe the arithmetic of matrices, including equality and addition of matrices,
multiplication of a matrix by a scalar, and multiplication of matrices. In general, division is not
defined for matrices, although an inverse can be defined for some matrices (see Section 5.3).
c
2019 School of Mathematics and Statistics, UNSW Sydney
174 CHAPTER 5. MATRICES
Note.
2. We use Mmn to stand for the set of all m × n matrices, i.e., the set of all matrices with m
rows and n columns.
In general, we will assume that the entries in a matrix can be complex numbers. However, it
is sometimes necessary to distinguish between real matrices, in which all entries aij are real
numbers, and complex matrices, in which all aij are complex numbers. In this case, we use
Mmn (R) for the set of all real m × n matrices and Mmn (C) for the set of all complex m × n
matrices. Likewise, Mmn (Q) is used for the set of all rational m × n matrices.
4. When we say “let A = (aij )”, we are specifying a matrix of fixed size, in which, for each given
i, j, the ijth entry is aij . On the other hand, for a given matrix A, we denote the entry in
the ith row and jth column by [A]ij .
1 2 −1 2
Example 1. The matrix A = 3 −4 2 5 is of size 3 × 4 and [A]24 = 5. ♦
−6 −3 1 4
When the number of rows or columns of a matrix is larger than 9, we need to use a comma
to separate the row number and the column number. If B is a matrix of size 5 × 12, we shall use
[B]2,4 and [B]3,11 to denote the entry in the second row and fourth column and the entry in the
third row and the eleventh column, respectively.
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.1. MATRIX ARITHMETIC AND ALGEBRA 175
To summarise, equal matrices have the same size and their corresponding entries are equal.
Addition of matrices is defined as follows.
To understand the proofs of the properties of matrices, we need to know the notation [A]ij well.
The row i column j entry of the matrix A + B is denoted by [A + B]ij . The definition simply
says that this entry is the same as [A]ij + [B]ij , that is the sum of the corresponding entries of the
matrices A and B.
Note that addition of matrices of different sizes is not defined.
Example 2.
2 3 5 −1 1 4 1 4 9
+ =
4 −3 2 2 4 −1 6 1 1
but
−1 1 4 1 2
+
2 4 −1 1 0
is not defined as the sizes are different. ♦
Hence (A + B) + C = A + (B + C).
c
2019 School of Mathematics and Statistics, UNSW Sydney
176 CHAPTER 5. MATRICES
Example 3.
0 0
0= is the zero matrix in M22 .
0 0
and
0 0 0
0= is the zero matrix in M23 .
0 0 0
♦
Note that when writing the zero matrix by hand, you can write 0 to distinquish it from the zero
vector and the number 0.
Proposition 2. Let A be a matrix and 0 be the zero matrix, both in Mmn . Then
A + 0 = 0 + A = A.
Proof. Similar to the proof of the associative law of addition but skipping the detail explanation,
we have
[A + 0]ij = [A]ij + 0 = [A]ij .
Hence A + 0 = A. We can also prove that 0 + A = A in a similar way.
Using the above definition of the negative of a matrix, we can define subtraction by
As with addition, we cannot subtract one matrix from another one of different size.
1 3 −2 2 −4 5
Example 4. Let A = and B = . We have
2 −1 3 3 0 −2
−2 4 −5 −1 7 −7
−B = and A − B = .
−3 0 2 −1 −1 5
A + (−A) = (−A) + A = 0.
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.1. MATRIX ARITHMETIC AND ALGEBRA 177
We now define multiplication of a matrix by a scalar, that is, by a number (real or complex).
The above rules for equality, matrix addition and multiplication by a scalar are essentially the
same as the corresponding rules for vectors in Rn . The matrix operations also obey the basic laws
of arithmetic as do addition and multiplication by a scalar in Rn . For instance, in both Mmn and
Rn , the commutative laws, associative laws and distributive laws hold.
It makes sense to use the above relation between the matrix equation Ax = b and the system of
linear equations as the basis for the definition of the product of two matrices. The column vector
x ∈ Rn may be regarded as an n × 1 matrix and the column vector b = Ax ∈ Rm may be regarded
as an m × 1 matrix.
c
2019 School of Mathematics and Statistics, UNSW Sydney
178 CHAPTER 5. MATRICES
To obtain a suitable general definition of multiplication of two matrices, we replace the matrix
x with 1 column by a matrix X with p columns. We can then think of the matrix A multiplying
each of the columns of X using the above rule.
The matrix X is the augmented matrix of the p column vectors, i.e. X = (x1 | · · · |xp ). So
the definition can be written as
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.1. MATRIX ARITHMETIC AND ALGEBRA 179
Note.
A B
m × n r × s
if equal
size of AB
Example 7. For
1 −1 2
1 2 3
A= and X = 2 −2 3
1 0 −1
3 −3 1
the product AX = (bij ) can be easily obtained.
1 −1 2
14 −14 11
A 2 = , A −2 = , A 3 = .
−2 2 1
3 −3 1
1 −1 2
1 2 3 2 −2 3 = 14 −14 11
Hence AX = .
1 0 −1 −2 2 1
3 −3 1
♦
Warning. For some A and X, AX 6= XA, i.e., matrix multiplication does not satisfy the commu-
tative law. This can happen when one product is defined and the other is not. It can happen even
when both are defined. Thus, for the A and X just given, the product XA is not defined as the
number of columns of X does not equal the number of rows of A.
X A
3 × 3 2 × 3
not equal
XA is not defined
c
2019 School of Mathematics and Statistics, UNSW Sydney
180 CHAPTER 5. MATRICES
Here is an example in which both AX and XA are defined but are different:
1 0 1 3
A= and X= ,
0 2 2 4
in which case
1 0 1 3 1 3
AX = = ,
0 2 2 4 4 8
whereas
1 3 1 0 1 6
XA = = .
2 4 0 2 2 8
A matrix is said to be square if it has the same number of rows as columns. For example, of
the four matrices displayed below, the second and fourth are square:
1 2 3
1 1 2 1 2 3 4 5 6
0 3 4 4 5 6
7 8 9
The diagonal of a square matrix consists of the positions on the line from the top left to the bottom
right. More precisely, the diagonal entries of an n × n square matrix (aij ) are a11 , a22 , . . . , ann .
The matrix is said to be an identity matrix if its diagonal entries are all 1 and all other entries
are 0. The following are identity matrices
1 0 0 0
1 0 0
1 0 0 1 0 0
(1) 0 1 0
0 1 0 0 1 0 .
0 0 1
0 0 0 1
For each integer n there is one and only one n × n identity matrix. It is denoted by In , or just by
I if there is no risk of ambiguity.
Definition 10. An identity matrix (written I) is a square matrix with 1’s on the
diagonal and 0’s off the diagonal.
Example 8.
1 0
I= is the identity matrix in M22 .
0 1
and
1 0 0
I = 0 1 0 is the identity matrix in M33 .
0 0 1
♦
Although matrix multiplication does not satisfy a commutative law of multiplication, the oper-
ation does satisfy some laws.
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.1. MATRIX ARITHMETIC AND ALGEBRA 181
Proof. We will prove the right distributive law, and leave the proof of the remainder as a problem
at the end of the chapter. As A + B exists, A and B must be the same size. Also, as AC exists,
the number of columns of A must be equal to the number of rows of C. Therefore we let A, B be
m × n matrices and C be an n × p matrix. Then BC also exists and the row i, column j entry of
(A + B)C is
n
X
[(A + B)C]ij = [A + B]ik [C]kj
k=1
Xn
= ([A]ik + [B]ik )[C]kj
k=1
Xn n
X
= [A]ik [C]kj + [B]ik [C]kj
k=1 k=1
= [AC]ij + [BC]ij
= [AC + BC]ij .
and
1 0 1 2 4
IA = = A.
0 1 −3 5 7
Note that in this case the right and left I’s must be of different sizes for the matrix products to
exist. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
182 CHAPTER 5. MATRICES
Example 10. Here is an example to verify the associative law of multiplication. For
−1 5
2 1 −2 −5 2
A= , B= 0 −3 , C= ,
3 4 2 −3 −6
2 6
we have
−10 −32
2 1 −2 45 18
A(BC) = 9 18 =
3 4 2 −50 −88
−28 −32
and
−6 −5 −5 2 45 18
(AB)C = = .
1 15 −3 −6 −50 −88
♦
Using the properties of matrix operations, we can simplify expressions in unknown matrices in
almost the same way as simplifying expressions in algebra.
Here A3 obviously means AAA. Thanks to the associative law of matrix multiplication, we do not
have to specify ABA to be (AB)A or A(BA) because (AB)A = A(BA). Note that we cannot write
the second term as 2A2 B because matrix multiplication is not commutative.
Solution.
(A + I)2 = (A + I)(A + I)
= (A + I)A + (A + I)I
= (A2 + IA) + (AI + I)
= A2 + A + A + I
= A2 + 2A + I
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.2. THE TRANSPOSE OF A MATRIX 183
a11 · · · a1n x1 b1 v1
.. . .. .. , x = .. , b = .. , v = ...
. . .
Proposition 6. Let A = . .
am1 · · · amn xn bn vn
v1
The vector v = ... is a solution to the matrix equation Ax = b, i.e. Av = b if and only if
vn
x1 = v1 , . . . , xn = vn is a solution to the system of equations
a11 x1 + · · · + a1n xn = b1
.. .. .. .
. . .
am1 x1 + · · · + amn xn = bm
Using this we can rewrite the proof of the propositions in Section 4.7 in a much shorter way.
As an example, let us rewrite the proof of the following.
Proposition 2 (Section 4.7). If v and w are solutions of Ax = 0 then so are v + w and λv for
any scalar λ.
A(v + w) = Av + Aw = 0 + 0 = 0.
A(λv) = λ(Av) = λ0 = 0.
c
2019 School of Mathematics and Statistics, UNSW Sydney
184 CHAPTER 5. MATRICES
Note. The columns of AT are the rows of A and the rows of AT are the columns of A.
Products of Vectors. So far we have not defined the product of two vectors. However, the
product of a row vector and a column vector makes sense as a special case of matrix multiplication.
Example 4. Let a and b be two vectors in Rn . Then, using the usual rule for matrix multiplication,
we have
b1
b2
aT b = a1 a2 . . . an . = a1 b1 + a2 b2 + · · · + an bn .
..
bn
This product is a 1 × 1 matrix1 , which we sometimes regard simply as a scalar, and so aT b is often
called the scalar product of the two vectors a and b in Rn . Note also that bT a = aT b. ♦
Example 5. If
1 2
4 −3
a=
−2 and b=
0
3 6
then
2 1
−3 4
aT b = 1 4 −2 3
0 = 2 − 12 + 0 + 18 = 8 = 2 −3 0 6 T
−2 = b a.
6 3
1 T
Technically, we should write the matrix product
a b as a matrix
with
brackets. In practice we write the result
1 2
as a single number without brackets.Thus if a = and b = , then we write aT b = 8, not aT b = (8).
2 3
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.2. THE TRANSPOSE OF A MATRIX 185
The order of multiplication is very important. If a and b are (column) vectors, then the product
abT is not a scalar but a matrix. ♦
Example 7. If
1 2
4 −3
a=
−2 and b=
0
3 6
then
2 −3 0 6
8 −12 0 24
abT =
−4
.
6 0 −12
6 −9 0 18
♦
Note that the expressions ab, aT bT have no meaning. The expression baT has a meaning as a
matrix. However, baT 6= abT in general.
Note that the matrix in Example 8 is the transpose of the matrix in Example 7. ♦
Proof. A transpose is obtained by changing the order of the subscripts on each entry, and hence
Thus all matrix entries of (AT )T are equal to corresponding entries of A, and hence the matrices
are equal.
c
2019 School of Mathematics and Statistics, UNSW Sydney
186 CHAPTER 5. MATRICES
Example 9.
T
T !T 1 4
1 2 3 1 2 3
= 2 5 = .
4 5 6 4 5 6
3 6
♦
Thus, as corresponding matrix entries are equal, we have (λA + µB)T = λAT + µB T , and the
result is proved.
Now, by definition of transpose, BT ∈ Mpn and AT ∈ Mnm . Thus the product B T AT exists. The
matrix entries of this product are
n
X n
X n
X
T T T T
[B A ]ij = [B ]ik [A ]kj = [B]ki [A]jk = [A]jk [B]ki = [(AB)T ]ij .
k=1 k=1 k=1
The result is proved.
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.2. THE TRANSPOSE OF A MATRIX 187
and
−2 4 1 1 3 5 −12
B T AT = 5 −1 3 2 −3 = 0 36 .
−1 5 2 −1 6 7 −6
♦
Example 11. For column vectors, a, b, we have from Propositions 1 and 3 that
Note that the entries of a symmetric matrix A satisfy [A]ij = [AT ]ji = [A]ji for all i, j, and
hence a symmetric matrix must be square.
Example 13. Let A and B be symmetric matrices. Prove that A + B is symmetric but AB may
not.
(A + B)T = AT + B T = A + B.
Hence A + B is symmetric.
2 1 1 2
Let A = and B = . Both A and B are symmetric. However,
1 0 2 0
4 4
AB =
1 2
is not symmetric.
c
2019 School of Mathematics and Statistics, UNSW Sydney
188 CHAPTER 5. MATRICES
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.3. THE INVERSE OF A MATRIX 189
Note.
1. The conditions in the definition of an invertible matrix are very restrictive, so many square
matrices are not invertible.
2. However, invertible matrices are very important as we shall begin to see below.
3. If A has left inverse then c 6 r. If A has a right inverse then r 6 c. (See proof of Proposition 1).
Notation: Because the inverse of a matrix A (if one exists) is unique we can denote it by a special
symbol: A−1 (read ‘A inverse’).
Proposition 3. If A is an invertible matrix, then A−1 is also an invertible matrix and the inverse
of A−1 is A. That is, (A−1 )−1 = A.
Proof. If X = A−1 exists, then AX = I and XA = I. But these two conditions also imply that A
is the inverse of X, and hence X is invertible and X −1 = A. Replacing X by A−1 then gives the
result (A−1 )−1 = A.
Proposition 4. If A and B are invertible matrices and the product AB exists, then AB is also an
invertible matrix and (AB)−1 = B −1 A−1 .
Thus X = B −1 A−1 also satisfies X(AB) = I. Thus, this X is the inverse of AB, and we have
B −1 A−1 = X = (AB)−1 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
190 CHAPTER 5. MATRICES
This proposition can easily be extended to products with 3 or more factors to obtain, for
example,
Example 2. Assuming F , G, and H are invertible and all products exist, simplify
A = HG(F HG)−1 F G.
A = HGG−1 H −1 F −1 F G.
A = HIH −1 IG = HH −1 G = IG = G.
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.3. THE INVERSE OF A MATRIX 191
−4 3 −4 3
Hence x1 = and x2 = and A−1 = . ♦
3 −2 3 −2
In general, if A is an invertible n × n matrix and the columns of the unique inverse, A−1 , are
x1 , x2 , . . . , xn , then xi is the unique solution of Ax = ei , where 1 6 i 6 n. Here {e1 , e2 , . . . , en }
is the standard basis for Rn . Conversely if any equation Ax = ei does not have a unique solution,
then A is not invertible.
This suggests a method of finding the inverse of a (invertible) matrix A:
1. Form the augmented matrix (A | I) with n rows and 2n columns.
3. Otherwise, use further row operations to reduce (U | C) to reduced row-echelon form (I | B).
The right hand half of this reduced row-echelon form is the inverse.
The next proposition sums up the implications of these observations.
Proposition 5. A matrix A is invertible if and only if it can be reduced by elementary row
operations to an identity matrix I and if (A | I) can be reduced to (I | B) then B = A−1 .
Example 4. Determine if the matrix
1 2 3
A= 2 1 2
1 −1 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
192 CHAPTER 5. MATRICES
− 21 5
6 − 61
A−1 = X = 0 1
3 − 32 .
1
2 − 12 1
2
Example 5. Determine if
1 2 3
A = 4 5 6
7 8 9
Here, all entries in the bottom row of U are zero, so A is not invertible. ♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.3. THE INVERSE OF A MATRIX 193
In this subsection we shall see that the three elementary row operations on a matrix A in Chapter 4
can be interpreted as multiplication of A by “elementary matrices”. As applications, we can use
this to prove that the algorithm to calculate the inverse of a matrix works and prove that the left
inverse of a square matrix is two-sided in the next subsection.
1 0 0 a b c a b c
Example 7. 1. If E1 = 0 3 0 and A = d e f then E1 A = 3d 3e 3f
0 0 1 g h i g h i
(check this), that is, the second row has been multiplied by 3.
1 0 0 0 a b c d a b c d
0 1 0 0 e f g h e f g h
2. If E2 =
0 0 0 1 and B = i j k l then E2 C = m n o p , that is, the
0 0 1 0 m n o p i j k l
3rd and 4th rows have been swapped.
1 0 0 a b c
3. If E3 = 0 1 0 then E3 A = d e f , that is, row three is replaced
−5 0 1 g − 5a h − 5c i − 5c
by row three minus five times row one. ♦
Proposition 6. The three elementary row operations can be effected by left multiplication by
matrices. These matrices are all invertible.
1
..
.
1
1
.. 0 1
.
1
Proof. Let E1 = λ E2 =
..
.
. ..
1
1 1 0
↑
..
i .
1
↑ ↑
1
i
j
..
.
1
..
E3 = . . The entries in the blank space are zeroes.
λ 1
..
.
1
↑ ↑
c
i
2019 School
j of Mathematics and Statistics, UNSW Sydney
194 CHAPTER 5. MATRICES
As demonstrated by Example 7 it is easy to see that, when multiplied on the left of a matrix,
• E1 multiplies row i by λ
• E2 interchanges row i and row j
• E3 adds λ times row i to row j
If is also clear that
1
..
1 .
..
. 1
..
E1−1 =
λ−1 , E −1 = E2 , E −1
2 3 = . .
..
. −λ 1
..
1 .
1
↑ ↑
i
j
Indeed E1−1 , E2−1 and E3−1 “undo” the operations done by E1 , E2 and E3 .
Corollary 9. The algorithm described in 5.3.2 does work. That is, the matrix calculated by the
algorithm is the inverse of the given matrix.
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.3. THE INVERSE OF A MATRIX 195
Proposition 11. Let A be an n × n square matrix. Then A is invertible if and only if the matrix
equation Ax = b has a unique solution for all vectors b ∈ Rn . In this case, the unique solution is
x = A−1 b.
Proof. Suppose A is invertible and Ax = b then A−1 Ax = A−1 b, Ix = A−1 b, x = A−1 b.
Conversely suppose Ax = b has a unique solution for all vectors b. Then, in particular, Ax = 0
has a unique solution, and this is clearly x = 0. Hence, by Proposition 10, A is invertible.
c
2019 School of Mathematics and Statistics, UNSW Sydney
196 CHAPTER 5. MATRICES
5.4 Determinants
In this section we shall define the determinant function. If A is a square matrix, then the determi-
nant of A, written det(A), is a number. Determinants arise in many areas of mathematics. As we
shall see in Chapter 5, they can be used to find the volume of a parallelepiped. Determinants are
very important in the theory of eigenvalues and eigenvectors, covered in Chapter 8. In later years
you will need to calculate determinants to perform changes of variables in multivariable calculus.
From a theoretical point of view, the determinant is also important in determining whether
a system of n linear equations in n unknowns has a solution for every right-hand-side vector, or
equivalently, whether an n × n matrix A is invertible. As we shall see, A is invertible if and only if
det(A) 6= 0. Unfortunately, calculating det(A), even by the most efficient methods, takes the same
length of time as finding the row-echelon form for A, so determinants are not used for numerical
calculations of this kind. On the other hand, many general statements can be proved by using the
properties of the determinant function.
Determinants are defined only for square matrices. We shall give definitions for arbitrary-
size n × n matrices, but most of our examples will be restricted to determinants of 2 × 2 and 3 × 3
matrices. The case of 1 × 1 matrices is included for completeness. An alternative notation for
det(A) is |A|.
Definition 1. Thedeterminant
of a 2 × 2 matrix
a11 a12
A= is det(A) = a11 a22 − a12 a21 .
a21 a22
2 −4
Example 1. If A = then |A| = det(A) = 2(4) − (−4)(−3) = −4. ♦
−3 4
The definition of the determinant of a general n × n matrix can be built up recursively from the
definition of the determinant of a 2 × 2 matrix.
We first define the concept of a minor of an entry of a matrix.
Notation. We shall use the symbol |Aij | to represent the (row i, column j) minor in a matrix A.
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.4. DETERMINANTS 197
2 −3 4
Example 2. The (row 2, column 3) minor of the (row 2, column 3) entry in A = 8 −1 −4
6 5 −7
2 −3
is, on deleting row 2 and column 3, |A23 | = = 28. ♦
6 5
Note that each term in the definition of the determinant |A| is a product of an entry in the first
row of A with its corresponding minor. The signs of the terms alternate +, −, +, −, . . . starting
with a + on the (row 1, column 1) entry. The formula is called “expanding along the first row of
the determinant.”
Solution.
3 −4 −2 −4
det(A) = 5 −1 + 7 −2 3 = 5(6 − 4) − (−4 + 24) + 7(2 − 18) = −122.
−1 2 6 2 6 −1
c
2019 School of Mathematics and Statistics, UNSW Sydney
198 CHAPTER 5. MATRICES
a22 a23 a21 a23 a21 a22
det(A) = a11 − a12
a31 a33 + a13 a31 a32
a32 a33
= a11 a22 a33 − a11 a23 a32 − a12 a21 a33 + a12 a23 a31 + a13 a21 a32 − a13 a22 a31
and
a22 a32 a12 a32 a12 a22
det(A ) = a11
T − a21
a13 a33 + a31 a13 a23
a23 a33
= a11 a22 a33 − a11 a32 a23 − a21 a12 a33 + a21 a32 a13 + a31 a12 a23 − a31 a22 a13
Example 4. Use expansion along the first column to evaluate the determinant of Example 3.
Solution. Expanding along the first column, we have
3 −4 1 7 1 7
det(A) = 5
− (−2) +6 = 5(6 − 4) + 2(2 + 7) + 6(−4 − 21) = −122.
−1 2 −1 2 3 −4
♦
Proposition 2. If any two rows (or any two columns) of A are interchanged, then the sign of
the determinant is reversed. More precisely if the matrix B is obtained from the matrix A by
interchanging two rows (or columns), then detB = −detA.
Proof. For 2 × 2 case. Let
a11 a12
A= .
a21 a22
Then, on interchanging columns 1 and 2 of A,
a12 a11
det = a12 a21 − a11 a22 = − det(A).
a22 a21
One important application of this proposition is that a determinant can be evaluated by ex-
panding along any row or any column, with the signs chosen from the following array.
+ − + − ···
− + − + · · ·
+ − + − · · ·
.
− + − + · · ·
.. .. .. .. . .
. . . . .
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.4. DETERMINANTS 199
If we evaluate the determinant by expanding the ith row or the jth column, where 1 6 i, j 6 n,
we have
n
X Xn
i+k
|A| = (−1) aik |Aik | = (−1)k+j akj |Akj |.
k=1 k=1
Example 5. Use expansion along the second row to evaluate det(A) of the previous two examples.
Solution. To evaluate det(A) by expanding along the second row, we choose the signs from the
+ − +
second row of − + − . So
+ − +
1 7 5 7 5 1
det(A) = −(−2)
+ 3
− (−4) = −122.
−1 2 6 2 6 −1
♦
Obviously, we should evaluate a determinant by expanding along that row or column containing
the greatest number of 0′ s. We will soon see an even better method.
A second application of Proposition 2 is as follows.
Proposition 3. If a matrix contains a zero row or column then its determinant is zero.
Proof. Clearly, if we evaluate the determinant by expansion along the zero row or column, the value
obtained will be zero.
Most of the remaining important properties of determinants centre around the question of what
happens to the value of the determinant when one column (or row) of a matrix is changed, with
all other columns (or rows) remaining unchanged.
Proposition 4. If a row (or column) of A is multiplied by a scalar, then the value of detA is
multiplied by the same scalar. That is, if the matrix B is obtained from the matrix A by multiplying
a row (or column) of A by the scalar λ, then detB = λdetA.
Proof. On multiplying the first row of
a11 a12 · · · a1n
a21 a22 · · · a2n
A= . .. .. ..
.. . . .
an1 an2 · · · ann
by a scalar λ, we obtain
λa11 λa12 · · · λa1n
a21 a22 · · · a2n
B= . .. .. .. .
.. . . .
an1 an2 · · · ann
Then, on expanding along the first row, we have
n
X n
X
det(B) = (−1)1+k λa1k |A1k | = λ (−1)1+k a1k |A1k | = λ det(A),
k=1 k=1
and hence the result is proved for a multiple of the first row. Then, from Propositions 1 and 2, the
result is also true for a scalar multiple of any row or column.
c
2019 School of Mathematics and Statistics, UNSW Sydney
200 CHAPTER 5. MATRICES
Example 6.
1 2 3
2 −5 7 =0
−3 −6 −9
as row 3 is a scalar multiple of row 1. This result can easily be checked directly by using expansion
along the second row to evaluate the determinant. ♦
Propositions 2 and 4 show the effect of the two elementary row operations of interchanging two
rows and of multiplying a row by a scalar. The effect of the third elementary row operation of
adding a multiple of one row to another row is given in the following proposition.
Proposition 6. If a multiple of one row (or column) is added to another row (or column), then
the value of the determinant is not changed.
[X] Proof. On adding λ times row i to row 1 of
a11 + λ ai1 a12 + λ ai2 · · · a1n + λ ain
.. .. .. ..
a11 a12 · · · a1n . . . .
.. .. .. .. ,
A= . we obtain
B= ai1 ai2 ··· ain .
. . .
.. .. .. .
..
an1 an2 · · · ann . . .
an1 an2 ··· ann
Then, on expanding along the first row, we have
n
X
det(B) = (−1)1+k (a1k + λ aik ) |A1k |
k=1
n
X n
X
1+k
= (−1) a1k |A1k | + λ (−1)1+k aik |A1k |
k=1 k=1
n
X
= det(A) + λ (−1)1+k aik |A1k | .
k=1
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.4. DETERMINANTS 201
The last sum is the determinant of the matrix which is obtained from A by replacing the first row
of A by the ith row of A. Thus, the first and ith rows of this matrix are the same, and hence, from
Proposition 5, its determinant is zero. Thus the second sum in the above equation is zero, and
hence det(B) = det(A). We have therefore proved the result for subtraction of scalar multiples of
any row from the first row. Then, from Propositions 1 and 2, the result is true for any two rows or
any two columns.
The final useful proposition that we shall give here is as follows.
Proposition 7. If A and B are square matrices such that the product AB exists, then
det(AB) = det(A) det(B).
All known proofs of this result are non–trivial, so we shall not give a proof here.
Example 7. For
2 −4 −1 −8
A= and B= ,
−3 5 2 9
we have
−10 −52
|A| = −2, |B| = 7, |AB| = = −14 = |A||B|.
13 69
♦
Proposition 8. If U is a square row-echelon matrix, then det(U ) is equal to the product of the
diagonal entries of U .
Proof. Let
u11 u12 · · · u1n
0 u22 · · · u2n
U = . .. .. ..
.. . . .
0 0 · · · unn
be an n × n matrix in row-echelon form. Then the proposition states that
c
2019 School of Mathematics and Statistics, UNSW Sydney
202 CHAPTER 5. MATRICES
Note that U (j) is a square row-echelon form matrix for 1 6 j 6 n, with U (1) = [u11 ], det(U (1)) =
u11 and U (n) = U .
We shall now prove that if the proposition is true for det U (j) , then it is also true for det U (j +
1) . Now,
u11 u12 · · · u1,j+1
0 u22 · · · u2,j+1
det U (j + 1) = . .. .. .. .
.. . . .
0 0 · · · uj+1,j+1
We can evaluate det U (j + 1) by expanding along the last row, and we obtain
det U (j + 1) = (−1)2j uj+1,j+1 det U (j) = uj+1,j+1 det U (j) .
If we now assume that the proposition is true for det U (j) , we can replace det U (j) by the
product of its diagonal entries to obtain
det U (j + 1) = u11 · · · ujj uj+1,j+1 .
Hence, if the result is true for det U (j) it is also true for det U (j + 1) .
However, the result is true for j = 1, since in this case we have det U (1) = u11 . It is therefore
true for j = 2, 3, . . ., and the proposition is proven for all n > 1 by induction.
Example 8.
3 2 9 3 12
0 −4 −5 6 2
0 0 6 −7 −3 = (3)(−4)(6)(5)(8) = −2880.
0 0 0 5 20
0 0 0 0 8
♦
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.4. DETERMINANTS 203
Proof. From Proposition 2, each row interchange reverses the sign of the determinant. Thus the
sign is unchanged if 0, 2, 4, . . . row interchanges have been made, whereas the sign is reversed if
1, 3, 5 . . . row interchanges have been made.
From Proposition 6, the value of the determinant is unchanged if a multiple of one row is added
to another row. Thus row addition operations do not change the value of the determinant. The
proof is complete.
Propositions 8 and 9 show that the value of a determinant can be found by the highly efficient
Gaussian elimination method.
Example 9. Evaluate the determinant of
1 −1 0 3
2 −2 6 −1
A=
4 −2
.
1 7
3 5 −7 2
Solution. The first leading element for Gaussian elimination is already in the first row, so no row
interchange is required. Hence, at this stage, ǫ = 1. Adding multiples of the first row to subsequent
rows does not change the value of the determinant. As is usual in Gaussian elimination, the multiple
are chosen to make the entries below the leading element all zero. After partly reducing the original
matrix we have:
1 −1 0 3
0 0 6 −7
|A| = .
0 2 1 −5
0 8 −7 −7
The second leading element is in row 3. We therefore interchange rows 2 and 3 to bring the
second pivot element to its required row 2 position. This interchange changes the sign of the
determinant, and hence we now have ǫ = −1. After interchanging rows 2 and 3 and then adding
suitable multiples of the new row 2 from the new row 3 and row 4, we have
1 −1 0 3
0 2 1 −5
|A| = (−1) .
0 0 6 −7
0 0 −11 13
The third pivot element is in its correct row 3 position, so no row interchange is required. Hence
ǫ remains at its previous value of −1. After further reducing the matrix we have
1 −1 0 3
0 2 1 −5 1
|A| = (−1)
0 0 6 −7 = (−1)(1 × 2 × 6 × 6 ) = −2.
0 0 0 1
6
The experienced evaluator of determinants will discard entries in these matrices as the calculation
progresses. This leads to the shortened calculation:
1 −1 0 3 1 −1 0 3
2 1 −5
2 −2 6 −1 0 0 6 −7 6 −7
= = (−1) 0
6 −7 = −2 = −2.
4 −2 1 7 0 2 1 −5 −11 13
0 −11 13
3 5 −7 2 0 8 −7 −7
c
2019 School of Mathematics and Statistics, UNSW Sydney
204 CHAPTER 5. MATRICES
♦
Although we can reduce any matrix to a row echelon form by the two operations—row exchange
and row addition, using the row operation—multiply a row by a scalar helps, especially when we
do the calculations by hand. This row operation exploits the result of Proposition 4. We illustrate
the technique by the following example.
1 a2 a3
[X] Example 10. Factorise 1 b2 b3 .
1 c2 c3
Solution.
1 a2 a3
1 b2 b3
1 c2 c3
1 a2 a3
R2 = R2 − R 1
= 0 b − a b − a
2 2 3 3
0 R3 = R3 − R 1
c 2 − a2 c 3 − a3
1 a2 a3 Proposition 4
= (b − a)(c − a) 0 a + b a2 + ab + b2
factorise (b − a) from R2
0 a+c a2 + ac + c2 factorise (c − a) from R3
1 a2 a3
= (b − a)(c − a) 0 a + b a2 + ab + b2
R3 = R3 − R 2
0 c−b a(c − b) + c2 − b2
1 a2 a3
Proposition 4
= (b − a)(c − a)(c − b) 0 a + b a + ab + b2
2
0 factorise (c − b) from R3
1 a+b+c
= (b − a)(c − a)(c − b)[(a + b)(a + b + c) − (a2 + ab + b2 )] expand along the first column
= (a − b)(b − c)(c − a)(ab + bc + ca)
♦
1. If det(A) 6= 0, the equation Ax = b has a solution and the solution is unique for all b ∈ Rn .
Proof.
CASE 1. det(A) 6= 0. From Propositions 8 and 9 , det(A) 6= 0 implies that all diagonal entries of
an equivalent row-echelon matrix U for A are non-zero. Then, since A, and hence U , is a square
c
2019 School of Mathematics and Statistics, UNSW Sydney
5.4. DETERMINANTS 205
matrix, U has no zero rows and no non-leading columns. As there are no zero rows, the equations
have a solution for all b, and, as there are no non-leading columns, each solution is unique.
CASE 2. det(A) = 0. In this case, U contains at least one zero diagonal entry, so at least one zero
row and at least one non-leading column. Thus there is a non-zero solution to Ax = 0. The result
follows from Proposition 5 of Section 4.7.
Note. In the case when det(A) 6= 0 there is an explicit formula called Cramer’s rule, available for
the solution of Ax = b. Each entry xi of x is specified as the quotient of two determinants.
Example 11. Describe the type of solution for each of the following systems.
x1 − x2 + 3x3 = b1
2x1 + 3x2 = b1
and 2x1 + 3x2 + x3 = b2 .
4x1 + 6x2 = b2
3x1 + x2 + 4x3 = b3
Solution. The determinants of the coefficient matrices of the equations are
1 −1 3
2 3
|A| = =0
and |B| = 2 3 1 = −5 6= 0.
4 6 3 1 4
Hence, depending on the values of b1 and b2 , the first system of equations either has no solution or
an infinite number of solutions, whereas the second system of equations has a unique solution for
all b. ♦
Note. With the possible exception of 2 × 2 and 3 × 3 matrices, Proposition 10 does not provide
a practical method of determining the number of solutions of Ax = b, because the most efficient
method for evaluating det(A) is first to use Gaussian elimination to reduce A to row-echelon form
U and then to multiply diagonal entries of U . However, as we have seen before, the number of
solutions that an equation Ax = b has can be seen directly from the row-echelon form itself,
without the need for any further calculation.
Proposition 10 leads immediately to two results which are sometimes useful.
Proposition 11. For a square matrix A, the homogeneous system of equations Ax = 0 has a
non-zero solution if and only if det(A) = 0.
Proof. Ax = 0 always has x = 0 as a solution. From Proposition 10, if det(A) 6= 0, then x = 0 is the
unique solution, whereas, if det(A) = 0, then there are an infinite number of non-zero solutions.
c
2019 School of Mathematics and Statistics, UNSW Sydney
206 CHAPTER 5. MATRICES
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 5. MATRICES 207
Find the following matrices if they exist, or explain why they don’t exist. (I stands for an
identity matrix of the appropriate size).
a) 3A, b) −2B, c) A + B, d) B + C, e) A + 3I,
f) B + 3I, g) AB, h) BA, i) BC, j) CD,
k) A2 , l) B 2 , m) (BD)2 .
2. [H] Suppose A and B are matrices such that both AB and BA are defined.
a) Show that AB and BA are both square matrices.
b) If AB = BA, show that A and B are both square and of the same size.
c) If A and B are square matrices such that AB = BA, show that
(A − B)(A + B) = A2 − B 2 .
d) Find two 2 × 2 matrices A, B for which (A − B)(A + B) 6= A2 − B 2 .
e) Prove that (A + B)2 = A2 + B 2 + 2AB if and only if AB = BA.
3. [H] Let A and B be matrices of the same size. By considering the general entries [A]ij , [B]ij ,
[A + B]ij and [B + A]ij , prove the commutative laws of addition, i.e. A + B = B + A.
5. [H] Let A and B be two matrices such that AB is defined. By considering the general entry
in both sides of the equation, show that A(λB) = λAB where λ is any real number.
6. [R] Let
1 0 1 1 2 2 2
A = 0 1 1 , B = 2 −2 , C = 3 −2 .
1 1 2 −1 4 −2 4
Show that AB = AC and deduce that matrices cannot in general be cancelled from
products.
7. [R][V] Let
2 1
A= .
3 −1
Show that A2 = A + 5I and hence find A6 as a linear combination of A and I.
c
2019 School of Mathematics and Statistics, UNSW Sydney
208 PROBLEMS FOR CHAPTER 5
8. [R] Let
0 1 0
N = 0 0 1 .
0 0 0
Find N 2 and N 3 . Show that (I + N ) (I − N + N 2 ) = I.
9. [H][V] Let A and B be n × n real matrices such that A2 = I, B 2 = I and (AB)2 = I. Prove that
AB = BA.
10. [H] Let A be a 2 × 2 real matrix such that AX = XA for all 2 × 2 real matrices X. Show that
A = αI for some α ∈ R.
14. [R] Let a = (1, 3, −2)T and b = (0, 4, 2)T . Evaluate all of the following expressions that make
sense and find those which are equal:
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 5. MATRICES 209
d) [H] Can you show how to write any square matrix as the sum of a symmetric and a
skew symmetric matrix?
16. [H] Show by constructing an example that, in general, AT A 6= AAT , even if A is square.
λ 0
17. [H] Suppose there exists a real matrix G such that GGT
= where λ, µ ∈ R. Prove
0 µ
that λ and µ are non-negative. If λ = 45 and µ = 20 find an example of such a matrix G
with integer entries.
18. [X] Show that a matrix A ∈ Mmn is a symmetric matrix if and only if:
19. [R][V] Find the inverses of those of the following 2 × 2 matrices that have inverses.
2 7 −4 7 6 12 8 9 0 1
a) , b) , c) , d) , e) .
1 4 3 −5 3 6 3 4 1 7
20. [R] Use the matrix inversion algorithm of Section 5.3 to decide if the following matrices are
invertible, and find the inverses for those which are invertible.
1 3 −2 0 2 0 1 2 3 1 4 1
A = 0 −1 2 , B = 1 2 3 , C = 2 3 4 , D = 2 3 1 .
0 0 1 −1 4 −2 4 5 6 1 −7 −2
21. [H] Write down the inverse of each of the following matrices
1 0 0 0 1 0
a) 0 5 0 b) 0 0 3 .
0 0 6 −2 0 0
22. [R] Decide if the following matrices are invertible, and find the inverses for those that are
invertible.
1 2 −1 −1 1 1 0 1 1 2 −2 −7
1 2 −2 −2 3 3 1 5 3 14
A= , B = , C = 2 4 .
0 1 1 1 1 0 2 4 −1 −2 3 11
1 4 0 −1 0 −4 2 1 3 5 2 12
c
2019 School of Mathematics and Statistics, UNSW Sydney
210 PROBLEMS FOR CHAPTER 5
25. [H] a) Prove that (AT )−1 = (A−1 )T for any invertible matrix A.
b) If A, B, C are invertible matrices of the same size simplify
i) A−1 (BAT )T B, ii) AT (CAT )−1 C T .
1 −1 1
26. [R][V] Let A = 1 1 0 .
3 −2 2
c1
a) Calculate A−1 . b) Solve Ax = c for x, where c = c2 .
c3
27. [H] A square matrix Q is said to be an orthogonal matrix if it has the property that QT Q = I.
That is, QT = Q−1 . Show that the matrix
2 1 2
3 3 −3
2
Q= 3 −3
2 1
3
1 2 2
3 3 3
29. [X] A complex generalisation of Question 27 is the following. A square matrix Q is said to be
T
a unitary matrix if it has the property that Q Q = I, where Q is the matrix obtained from
Q by taking complex conjugates of each entry of Q. Give an example of a 2 × 2 unitary
matrix with non-real entries.
30. [X] Let Q be a square n × n orthogonal matrix, i.e., a square matrix for which QT Q = I,
where I is an identity matrix. Show that (Qx) · (Qy) = x · y for all x, y ∈ Rn .
HINT. x · y = xT y.
31. [X] Let Q be a square n × n orthogonal matrix. Show that the columns of Q are a set of n
orthonormal vectors in Rn . Show that the rows of Q also form a set of n orthonormal
vectors in Rn .
32. [X] Let Q be a square n × n orthogonal matrix. Let {e1 , e2 , . . . , en } be the n standard basis
column vectors of Rn . Show that the set of vectors {Qe1 , Qe2 , . . . , Qen } also form a set of
orthonormal vectors.
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 5. MATRICES 211
35. [R] Evaluate the determinants of the following 2 × 2 matrices and hence determine whether
or not they are invertible.
2 7 −4 7 5 2 8 9 11 13
a) , b) , c) , d) , e) .
1 4 3 −5 10 4 3 4 12 14
36. [R] Evaluate the determinants for the following matrices by reducing to row echelon form.
−1 1 2 1 −2 4 1 0 4
a) 2 4 −1 , b) 3 1 −2 , c) 3 1 −2 .
0 −1 1 1 5 −10 1 5 −10
1 1 1 1
1 1 7 1
37. [R] Find the determinant of the matrix
1
.
8 3 1
1 1 1 4
a b c
38. [H] Suppose A = d e f has determinant 5. Find
g h i
3a 3b 3c a + 2d b + 2e c + 2f
a) det 2d 2e 2f , b) det d − g e − h f − i ,
−g −h −i g h i
d e f
c) det g h i , d) det(7A).
a b c
c
2019 School of Mathematics and Statistics, UNSW Sydney
212 PROBLEMS FOR CHAPTER 5
42. [H] Long long ago, a mathematician wrote C and C −1 on a piece of paper. Unfortunately
insects have damaged the paper and all that is left is
−2 −1 1 ∗ 0 −1
C= 1 2 −1 and C −1 = 2 ∗ −1
∗ ∗ ∗ 5 1 ∗
44. [H] Let U1 and U2 be two n×n row-echelon matrices. Prove that det(U1 ) det(U2 ) = det(U1 U2 ).
α 1 −1
45. [H] Let A = α 2α + 2 α .
α−3 α−3 α−3
a) Factorise det(A).
b) Hence, find the values of α will there be a nonzero solution of Ax = 0.
46. [R] Show by constructing an example that in general det(A + B) 6= det(A) + det(B).
48. [H] Use the product rule for determinants to show that a square orthogonal matrix Q (see
Question 27) has a value for det(Q) of +1 or −1.
49. [X] Use the product rule for determinants to show that a square unitary matrix Q (see Ques-
tion 29) has det(Q) = eiθ for some angle θ.
50. [X] Let A and B be two matrices which differ only in the first column, i.e., let
A = a1 a2 · · · an and B = b 1 a2 · · · an ,
c
2019 School of Mathematics and Statistics, UNSW Sydney
CHAPTER 5. MATRICES 213
where a1 and b1 are the first columns of A and B and where ai , i = 2, 3, . . . , n, are the
remaining columns of both A and B. Let C be the matrix
C = a1 + b 1 a2 · · · an
obtained by replacing the first column of A (or B) by the sum of the first columns of A
and B.
Show that
det(C) = det(A) + det(B).
Explain why the result of this question also holds for adding two matrices which differ
only in one column (not necessarily the first) or which differ only in one row.
51. [X] Prove the following relationships between “volumes” and determinants:
a1 b1
a) In two dimensions, let a = and b = , and consider the parallelogram
a2 b2
spanned by a and b. Show that a parametric vector form for the parallelogram is
x = λ1 a + λ2 b for 0 6 λ1 6 1, 0 6 λ2 6 1,
and then show that the area of the parallelogram is equal to | det(A)|, where A is the
matrix with rows a and b.
a1 b1 c1
b) In three dimensions, let a = a2 , b = b2 and c = c2 , and consider the
a3 b3 c3
parallelepiped spanned by a, b and c. A parametric vector form for the equation of
the parallelepiped is
x = λ1 a + λ2 b + λ3 c for 0 6 λ1 6 1, 0 6 λ2 6 1, 0 6 λ3 6 1.
Show that the volume of the parallelepiped is equal to | det(A)|, where A is the matrix
with rows a, b and c.
c) What is the one-dimensional version of these results?
c
2019 School of Mathematics and Statistics, UNSW Sydney
214 PROBLEMS FOR CHAPTER 5
zx + y = 2, x + zy = 3, x + y = z + 1.
56. [X] Suppose α, β and γ are the roots of the cubic equation x3 +px+q = 0 and sk = αk +β k +γ k .
Find s1 , s2 , s3 in terms of p and q and show that
s1 s2 s3
det s2 s3 s1 = 8p3 + 27q 3 .
s3 s1 s2
57. [X] Let A(x1 , y1 ), B(x2 , y2 ), C(x3 , y3 ) be three points in the plane.
b) Now suppose that A, B, C are not collinear. By considering the areas of some trapezia
(or otherwise), show that the area of the triangle with vertices A, B, C is given by
|D| where
x 1 y1 1
2D = det x2 y2 1 .
x 3 y3 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS TO SELECTED
PROBLEMS
Chapter 1
1. a) a + h, b) a − h, c) a + 12 h, d) 3
4 a, e) 3
4 a − 21 h.
−→
2. a) 0, b) 2CA.
1
4. a) 2 (b + a), 21 (b + c)
14 cm
75◦ 13 cm/s 12 cm/s
5 cm
10 cm
8 × 13 = 104 cm
18 km/h
h
25 km/h m/
2k 1 km/h
15 km/h √
3 km/h
1 1 1 3 1 3 1 3 3
8. a) b + c, a + c, a + b. b) a + b + c.
2 2 4 4 4 4 7 7 7
−7
16
3 2
9. a) , b) 15 , c)
−6 , d) Not possible, e) 7i − 4j + 3k.
4
−5
−1
22. d + e − f , d + f − e, e + f − d.
23. The midpoint is (3, −1, 3). The point Q is (10, −29, 31).
1 2
24. t = a + b
3 3
1 0 0 0 0
0 1 0 0 0
25.
0 , 0 , 1 , 0 , 0 .
0 0 0 1 0
0 0 0 0 1
4
2
4 0
1 √ 1 1 √ 1
26. 6, −4 ; 14, √ ; 21, √ 1 .
6
2 14 0
21 −2
3
0
√
27. a) 15, b) 12, c) 62.
√ √ √
28. 35, 6, 41.
30. (5, 09 )T , (0, 5, 08 )T . . . , (09 , 5)T . Yes, (α, α, . . . , α)T where (5 − α)2 + 9α2 = 50.
1 −2
1 1
31. a) x = +λ , λ ∈ R; b) x = 2 + λ −3 , λ ∈ R;
2 5
−1 6
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 217
1 −2
1 6 2 1
c) x = 2 + λ 0 , λ ∈ R; d) x =
−1 + λ 2 , λ ∈ R.
1 2
3 −2
c
2019 School of Mathematics and Statistics, UNSW Sydney
218 CHAPTER 1
3 1
39. a) x = λ −1 + µ 4 ; λ, µ ∈ R.
2 −6
1 1 0
b) x = 4 + λ 2 + µ −14 ; λ, µ ∈ R.
−2 6 5
1 −2
40. a) Plane through the origin parallel to 2 and 3 .
3 4
−2
1
b) Line through (3, 1, 2, 4) parallel to
3 .
2
3
2
c) Line through origin parallel to
1 .
2
4 8
d) Plane through (1, 2, 3) parallel to −1 and 2 .
2 4
1 2 −1
41. a) x = 2 + λ1 1 + λ2 2 for λ1 , λ2 ∈ R;
3 3 −3
3 −4 3
b) x = 1 + λ1 1 + λ2 6 for λ1 , λ2 ∈ R;
4 0 −6
−2 5 3
4
c) x= + λ1 −2 + λ2 0 for λ1 , λ2 ∈ R;
1 5 −1
6 −7 −6
3 −3 3
d) x = 0 + λ1 0 + λ2 4 for λ1 , λ2 ∈ R;
0 2 0
0 1 0
e)
x = 1 + λ1 0 + λ2 65 for λ1 , λ2 ∈ R;
0 0 1
1 4 7
2 0 2
f) x=
3 + λ1 −4 + λ2 −3 for λ1 , λ2 ∈ R.
4 5 −5
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 219
−1 −1
42. a) x = λ1 1 + λ2 0 ; λ1 , λ2 ∈ R.
0 1
4 1 −4
b) x = 0 + λ1 3 + λ2 0 ; λ1 , λ2 ∈ R.
0 0 3
0 1 0
c) x = −1 + λ1 0 + λ2 −6 ; λ1 , λ2 ∈ R.
0 0 1
0 1 0
d) x = 0 + λ1 0 + λ2 1 ; λ1 , λ2 ∈ R.
2 0 0
48. a) Parallelogram with vertices (0, 1), (1, 3), (2, 4), (3, 6).
b) Triangle with vertices (0, 1), (1, 3), (3, 6).
c) Parallelogram with vertices (0, 0, 0), (12, 6, −12), (32, −16, 24), (44, −10, 12).
d) Triangle with vertices (0, 0, 0), (12, 6, −12), (36, −6, 6).
e) An
unbounded
region with vertices O and
P and two of the three sides parallel to
4 36
−2 −−
→ −6
3 . At P , λ1 = λ2 = 6 and so OP = 6 .
−1 6
P
2
1
O
−2
2 4
−2
3
−1
c
2019 School of Mathematics and Statistics, UNSW Sydney
220 CHAPTER 2
1 −3 −2
3 −2 −1
b) x =
4 + λ1 −4 + λ2 6 for 0 6 λ1 6 1, 0 6 λ2 6 λ1 .
2 3 3
c) The three parallelograms are:
C(−4, 0, 6, 8) C C
A B
A(1, 3, 4, 2) B(−2, 1, 0, 5) A B
Chapter 2
π 1 π 7
1. a) b) cos−1 √ ≈ 86◦ 41′ , c) , d) cos−1 √ ≈ 78◦ 48′ .
4 10 3 2 10 13
2 1 4 5 8 7 1 8
2. a) 0, √ , √ ; b) √ , −√ , √ ; c) √ , −√ , √ .
6 3 3 2 33 66 3 10 42 105
1
3. cos−1 ≈ 70◦ 32′ .
3
1 3
7. λ1 = a · u1 = √ , λ2 = a · u2 = −3, λ3 = a · u 3 = √ .
2 2
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 221
5 √ 80
5 π 66 1
8. a) 2 . b) . c) . d) 50 .
2 2 17
1 22
−1
2 −3
1 3
3 ,
9. a) −4 , b) c) 3 .
3 14 0
2 6
2
√
10. a) 7, b) 3, c) 6.
(a · b)2
12. b) q(λ0 ) = a · a −
b·b
16 −23 −45
14. a) −4 , b) −11 , c) 9 .
−2 20 −18
12
15. −8 .
6
√ 8 √ 0
17. a) 2 21, −4 ; b) 2 2, −2 .
2 −2
√ 15
18. a) 2; b) .
2
4 1
19. a) − √ . b) √ .
3 2 2
1
20. a) 2, b) √ , c) 7.
2
1 −2
23. a) Line through A and B is x = 0 + λ1 2 , λ1 ∈ R.
2 1
0 2
Line through C and D is x = 1 + λ2 −1 , λ2 ∈ R.
1 −2
3
b) Shortest distance is √ .
17
21 38 53 30 32 47
c) Point P is − , , and Q is − , , .
17 17 17 17 17 17
c
2019 School of Mathematics and Statistics, UNSW Sydney
222 CHAPTER 2
27. As usual, the answers for equations of planes are not unique.
3 1 2 −1 1
a) x = 0 + λ1 1 + λ2 0 ; 1 · x − 2 = 0;
0 0 1 2 −2
x1 − x2 − 2x3 = 3.
1 −1 2 −5 1
b) x = 2 + λ1 1 + λ2 3 ; 5 · x − 2 = 0;
−2 2 1 −5 −2
x1 − x2 + x3 = −3.
1 −2 1 −7 1
c) x = 2 + λ1 −1 + λ2 1 , 10 · x − 2 = 0;
−2 4 3 −1 −2
7x1 − 10x2 + x3 = −15.
−1 1/2 −1/4 −4 −1
d) x = 0 + λ1 1 + λ2 0 , 2 · x − 0 = 0;
0 0 1 −1 0
4x1 − 2x2 + x3 = −4.
1 −1 14 4 1
e) x = 2 + λ1 2 + λ2 2 , 17 · x − 2 = 0;
−2 −2 −6 15 −2
4x1 + 17x2 + 15x3 = 8.
1 1 2
28. a) x = 2 + λ 0 + µ −1 , for λ, µ ∈ R.
4 −1 −1
−1
b) −1 . c) x1 + x2 + x3 = 7.
−1
4 5
1
29. a) 4 , b) 2 .
2
2 −1
√ 13 25
30. a) 3, b) 6, c) , d) .
7 7
1 −1 −2
31. a) x = 2 + λ −1 + µ 1 , λ, µ ∈ R.
0 2 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 223
1 1 1
8
b) 1 . c) 1 · x − 2 = 0.
d) √ .
1 1 0 3
3 8
11 1 11
12
32. a) c = proja v, d = v − c. b) c = − 11 , d = 11
.
1 12
− 11 11
Chapter 3
1.
x∈N x∈Z x∈Q x∈R
a) - −25 −25 −25
3 3 3 3
- −3 −3 −3
- - − 10
3 − 10
3
b) 1 1,−5 1,−5 1,−5
5 5 5, 32 5,√32
1± 5
- - - 2
- - - -
c) 3j,j ∈ N 3j,j ∈ Z 3j,j ∈ Z 3j,j ∈ Z
0 0 0 3kπ,k ∈ Z
2. No.
4. Yes.
z 1
5. 3z = 6 + 9i, z 2 = −5 + 12i, z + 2w = 7i, z(w + 3) = −2 + 10i, = (4 − 7i),
w 5
w 1
= (4 + 7i).
z 13
1 1
6. a) (3 − i), b) − (1 − i).
5 2
a b 1
7. a) a2 − b2 + 2abi, b) −i 2 , c) (a2 − 1 + b2 ) − 2ib .
a2 +b 2 a + b2 (a − 1)2 + b2
1
√ √ 1
√
8. a) 2 (−1 ± 3 i), b) −1 ± 2 i, c) 3 ± i, d) 2 (3 ± 3)i, e) ±i, ±2i.
10. 16
c
2019 School of Mathematics and Statistics, UNSW Sydney
224 CHAPTER 3
8abi(a2 − b2 )
11.
(a2 + b2 )2
12.
z Re(z) Im(z) z
−1 + i −1 1 −1 − i
2 + 3i 2 3 2 − 3i
2 − 3i 2 −3 2 + 3i
2−i 1
1+i 2 − 32 1+3i
2
1
(1+i)2
0 − 21 1
2
11 2
13. −3 + 4i, − i.
25 25
17. b) z 2 − 6z + 13
18.
z |z| Arg(z) Polar Form
√ π
√
6 + 6i 6 2 4 6 2 cos π4 + i sin π4
−4 4 π 4(cos π + i sin π)
√ −π
3−i 2 6 2 cos π6 − i sin π6
−1 √i −3π
√
2
− 2
1 4 cos 3π 3π
4 − i sin 4
√ √
−7 + 3i 58 α 58(cos α + i sin α) Here α = π − tan−1 73 .
√
19. 234, −1.
20. n = 4
3 √ 3 √ 3 √ 3 √
21. a) 1 + 3i , b) − 3+i , c) − 1 + 3i , d) 3−i ,
2 2 2 2
3 p √ p √
e) 2 + 2 + i 2 − 2 (Double angle formula used).
2
√ √
√ 1+ 3 3−1
27. 64, −(1 + 3)i, + i.
2 2
7
28. .
2
29. π.
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 225
3π √ 5π
30. Arg (−1 + i) = ; Arg − 3 + i = ;
4 6
√ 5π −1 + i π
Arg (−1 + i) − 3 + i = − ; Arg √ =− .
12 − 3+i 12
√
7π 1+ 3
31. sin = √ .
12 2 2
√ √ h π π i z 12
32. zw = 2 2eiπ/12 = 2 2 cos + i sin ; z 9 = −512; = 64eiπ = −64.
12 12 w
√
√ 1 3
33. a) 16 − 3 + i , b) − i, c) − − i .
2 2
1 iθ
53. a) cos θ = e + e−iθ .
2
1
b) cos6 θ = (cos 6θ + 6 cos 4θ + 15 cos 2θ + 10).
32
1
54. sin5 θ = (sin 5θ − 5 sin 3θ + 10 sin θ)
Z 16
5 1 1 5
sin θdθ = − cos 5θ + cos 3θ − 10 cos θ + C,
16 5 3
c
2019 School of Mathematics and Statistics, UNSW Sydney
226 CHAPTER 3
1
cos4 θ =[3 + 4 cos(2θ) + cos(4θ)]
Z 8
4 1 1
cos θdθ = 3θ + 2 sin(2θ) + sin(4θ) + C.
8 4
1
1
sin 2 (n + 1)θ sin 2 nθ
58. .
sin 12 θ
9eiθ
59. a) .
9 + ei2θ
π 2π
60. a) |z − i| 6 2 b) |z − i| 6 2 or − 6 Arg (z − i) 6
3 3
2
i b b
i
y=3
x
=
y
2 b
0
0
y = −3
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 227
π π
e) The real axis f) |z − 1 − i| < 1 & − < Arg (z − 1 − i) 6
4 2
Im
1+i
0
0 Re
2 2 y 2
5 2 4 x
g) Circle: x2 + y + = h) Ellipse: √ + =1
3 3 2 2 3
b 3i
0
√ b b
√
−2 2 0 2 2
b
− 53 i
b
b
−3i
−3i
Im Axis
x = 3y
Real Axis
0 3 S1
c
2019 School of Mathematics and Statistics, UNSW Sydney
228 CHAPTER 3
π π
b) |z − i| < |z + i| and − 6 Arg (z − i) 6
6 6
Im Axis
S2
π
6 |z − i| < |z + i| is equivalent to
1 π Im(z) > 0
6
0 Real Axis
Im(z)
3
1+i
Re(z)
Im(z) = −4
z
b
| z − Re(z) |
|z
−
x|
b b
0 Re(z) x
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 229
iα |
e z
|z − b
|
e iθ
b
1 α
−
|z
α θ
Real axis Real axis
√ √
69. a) (x − 1)(x + 1)(x2 + 1)(x2 + 2x + 1)(x2 − 2x + 1).
√ √
b) (x2 + 2)(x2 + 6x + 2)(x2 − 6x + 2).
70. (z 2 + 2z + 2)(z 2 − 2z + 2)
Im z
i b
5iπ b b
iπ
e 6 e6
Re z
0
−5iπ b b
−iπ
e 6 e 6
−i
5πi πi πi πi πi 5πi
− 6 −2 −6
c) z−e z−e z−e z−e 6 z−e 2 z−e 6 .
√ √
d) z 2 + 1 z 2 + 3 z + 1 z 2 − 3 z + 1 .
c
2019 School of Mathematics and Statistics, UNSW Sydney
230 CHAPTER 4
75. a) (t + 1 − i) (t + 1 + i) (t − 2) (t + 1) (t + i) (t − i),
b) (t2 + 2t + 2) (t − 2) (t + 1) (t2 + 1).
√ √
√
3
3
5 √ 3
5 √
76. 1 + i, 1 − i, 5, −1 + i 3 , −1 − i 3 .
2 2
√ π √ 7π
80. d) −2, 2 2 cos , 2 2 cos .
12 12
5 1 1
81. a) 1; b) −1, , ; c) 4, ± .
7 4 5
90. evalc((sqrt(2)+7*I)^13);
Chapter 4
5
5
2
5 2 λ : λ, µ ∈ R
1. a) , :λ∈R ,
2 λ
µ
4 − 2λ
4 − 2λ λ : λ, µ ∈ R
b) :λ∈R ,
λ
µ
λ
c) µ : λ, µ ∈ R
2 − 2λ + 3µ
x1
8
2. a) No solution. b) Unique solution = .
−9
x2
5 5
c) Infinite number of solutions on the line x = +λ , λ ∈ R.
0 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 231
8. a) In vector form,
3 −3 4 6
x1 5 + x2 2 + x3 −3 = 7 .
−1 −1 6 8
As a matrix equation and augmented matrix,
3 −3 4 x1 6 3 −3 4 6
5 2 −3 x2 = 7 ;
(A|b) = 5 2 −3 7 .
−1 −1 6 x3 8 −1 −1 6 8
b) In vector form,
1 3 7 8 −2
x1 3 + x2 2 + x3 −5 + x4 −1 = 7 .
0 3 6 −6 5
As a matrix equation and augmented matrix,
x1
1 3 7 8 −2 1 3 7 8 −2
x
3 2 −5 −1 2 = 7 ;
x3 (A|b) = 3 2 −5 −1 7 .
0 3 6 −6 5 0 3 6 −6 5
x4
c
2019 School of Mathematics and Statistics, UNSW Sydney
232 CHAPTER 4
x1 − 3x2 = 10
6x2 + 6x3 = −2
−6x1 − x2 − 4x3 = 0
7x1 + 9x2 + 11x3 = 5
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 233
1 0 0 −1
15. a) 0 1 0 2 .
0 0 1 −2
−1
Solution: x = 2 , which is the position vector of a point in R3 .
−2
1 0 0 −75 −34
b) 0 1 0 29 13 .
0 0 1 7 3
−34 75
13 −29
Solution: x =
3 + λ −7 , λ ∈ R, which is a line in R .
4
0 1
20. Perhaps, if the costs are negative or very large then you can be sure that someone is cheating.
21. No.
3
22. a) x1 = 7b1 + 5b2 + 3b3 b) x1 = 2 b1 − 2b2 − 2b3
x2 = 6b1 + 4b2 + 3b3 x2 = − 27 b1 + 5b2 + 4b3
x3 = 2b1 + b2 + b3 1
x3 = 2 b1 − b2 − b3
26. Yes.
27. No.
1 3 4
1
28. Yes, since = 3 −1 − 2 −2 .
4 4 4
12 6 3
29. No.
c
2019 School of Mathematics and Statistics, UNSW Sydney
234 CHAPTER 4
7 2 4 6
only when λ1 = λ2 = µ1 = µ2 = 0.
1
1 3
37. a) x = 0 + λ 23 , λ ∈ R. b) The planes intersect in a line.
0 1
41. 3, 1, 2.
42. a) Letting x1 be the number of hectares of wheat, x2 be the number of hectares of oats and
x3 be the number of hectares of barley gives the equations
x1 + x2 + x3 = 12
6x1 + 6x2 + 2x3 = 48
150x1 + 100x2 + 70x3 = 700
72x1 + 48x2 + 36x3 = 612
b) There is no solution.
c) The inequalities are
x1 + x2 + x3 6 12
6x1 + 6x2 + 2x3 6 48
150x1 + 100x2 + 70x3 6 700
72x1 + 48x2 + 36x3 6 612
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 235
x1 + x2 + x3 + s1 = 12
6x1 + 6x2 + 2x3 + s2 = 48
150x1 + 100x2 + 70x3 + s3 = 700
72x1 + 48x2 + 36x3 + s4 = 612
d) Some sensible solutions are to either plant 4 32 hectares wheat and no oats and barley, or
7 hectares oats and no wheat and barley, or 10 hectares barley and no wheat and oats.
There are also an infinite number of other reasonable solutions. In each case it is the
fertiliser which is restricting the planting.
44. a) Π1 is x + 2y − z = 2, Π2 is 3x + 6y − z = 12, Π3 is 2x + 4y − z = 7.
5 − 2t2 5 −2
b) x = t2 = 0 + t2 1 , t2 ∈ R
3 3 0
−2
The intersection is a line through (5, 0, 3) and parallel to 1 .
0
c) x = −2y + 5 and z = 3.
x − 2y + z = a
4 1 1
3x + 6y + 8z = b
45. a) . b) d − a + 2b − 3c = 0. c) ,− , .
4x + 2y + 7z = c 7 7 7
7x − 8y + 6z = d
Chapter 5
c
2019 School of Mathematics and Statistics, UNSW Sydney
236 CHAPTER 5
6 −9 12 4 −2
1. a) 3A = 9 6 −6 . b) −2B = −6 −8 .
3 −3 9 2 −10
−5 3
c) A + B is not defined. d) B + C = 4 0 .
5 7
5 −3 4
e) A + 3I = 3 5 −2 . f) B + 3I is not defined.
1 −1 6
−17 10
g) AB = 2 1 . h) BA is not defined.
−8 12
−4 −13 −9
i) BC is not defined. j) CD = −2 11 13 .
14 14 0
−1 −16 26
k) A2 = 10 −3 2 . l) B 2 is not defined.
2 −8 15
−86 81 167
m) (BD)2 = −47 38 85 .
−187 171 358
7. 96A + 205I.
0 0 1 0 0 0
8. N 2 = 0 0 0 , N 3 = 0 0 0 .
0 0 0 0 0 0
0 0
11. a) 1, b) 0 0 1 , c) −1 , d) 1 −2 3 .
0 0
2 −4 5
1 4 2
1 −3 4 −5 6 0
13. AT = , BT =
4
, C T = 4 −3 6 = C.
−2 0 5 5 8
2 6 7
3 5 6
0 4 2 0 0 0
14. aT b = bT a = 8, abT = 0 12 6 , baT = 4 12 −8 , ab and aT bT are not
0 −8 −4 2 6 −4
defined.
3 6
17. A possible G = .
−4 2
c
2019 School of Mathematics and Statistics, UNSW Sydney
ANSWERS 237
4 −7 5 7 1 4 −9 −7 1
19. a) , b) , c) no inverse, d) , e) .
−1 2 3 4 5 −3 8 1 0
1 3 −4 8 −2 −3
20. A−1 = 0 −1 2 , B −1 = 12 0 0 , C is not invertible,
0 0 1 −3 1 1
1 1 1
1
D−1 = 5 −3 1 .
4
−17 11 −5
1 0 0 0 0 − 21
21. a) 0 51 0 b) 1 0 0
1 1
0 0 6 0 3 0
4 −3 −2 0 6 −2 1 0
−1 1 1 0 9 −4 3 −1
22. A−1 =
1 −2 −2
; B −1 = ;
1 25 −11 8 −2
0 1 2 −1 −14 6 −4 1
C −1 does not exist.
2
23. a) B −1 , b) AB 6 A−1 , c) (A + A−1 )2 , d) I − (I − A)m+1 .
2 4 4
24. a) A−1 B. b) 1 −2 3 .
1 0 3
25. b) i) B T B, ii) C −1 C T .
−2 0 1 −2c1 + c3
26. a) 2 1 −1 . b) 2c1 + c2 − c3 .
5 1 −2 5c1 + c2 − 2c3
27. x = QT b.
!
√1 i − √12 i
2
29. e.g. .
√1 i √1 i
2 2
T
33. From Question 29, Q is invertible, and hence Qx = b has the solution x = Q−1 b = Q b.
1 b 0 A−1 0
34. a) . b) .
ab −c a −B −1 CA−1 B −1
5 2
35. a) 1, b) −1, c) 0, d) 5, e) −2. All are invertible except .
10 4
c
2019 School of Mathematics and Statistics, UNSW Sydney
238 CHAPTER 5
37. −126.
1
39. a) −2, b) − , c) −32.
2
40. −83, −108, 8964.
41. a 6= 1.
1 0 −1 −2 −1 1
42. a) 2 1 −1 . b) 1 2 −1 . c) 1.
5 1 −3 −3 −1 1
c
2019 School of Mathematics and Statistics, UNSW Sydney
240 INDEX
Index
c
2019 School of Mathematics and Statistics, UNSW Sydney
INDEX 241
orthogonal, 50
parallelepiped, 64
pivot, 142
plane, 29
Cartesian form, 65
parametric vector form, 32
point-normal form, 66
spanned by two vectors, 30
polynomial, 109
factorisation theorem, 110
root, 109
position vector, 15–17
projection, 54
scalar quantity, 1
scalar triple product, 62
span, 30
system of linear equations, 133
consistent, 133
homogeneous, 133
inconsistent, 133
solution, 133